SOLVENT RECYCLING, REMOVAL, AND DEGRADATION

21

Solvent Recycling, Removal, and Degradation 21.1 ABSORPTIVE SOLVENT RECOVERY Klaus-Dirk Henning CarboTech Aktivkohlen GmbH, Essen, Germany

21.1.1 INTRODUCTION A variety of waste gas cleaning processes1-4 is commercially available. They are used to remove organic vapors (solvents) from air emissions from various industries and product lines. The main applications of solvents in various industries are given in Table 21.1.1. Table 21.1.1. Application of organic solvents. [After references 2-4]. Application

Solvents

Coating plants: adhesive tapes, mag- Naphtha, toluene, alcohols, esters, ketones (cyclohexanone, methyl netic tapes, adhesive films, photo- ethyl ketone), dimethylformamide, tetrahydrofuran, chlorigraphic papers, adhesive labels nated hydrocarbons, xylene, dioxane Chemical, industries

pharmaceutical,

food Alcohols, aliphatic, aromatic and halogenated hydrocarbons, esters, ketones, ethers, glycol ether, dichloromethane

Fiber production Acetate fibers Viscose fibers Polyacrylonitrile

Acetone, ethanol, esters carbon disulfide dimethylformamide

Synthetic leather production

Acetone, alcohols, ethyl ether

Film manufacture

Alcohols, acetone, ether, dichloromethane

Painting shops: cellophane, plastic Amyl acetate, ethyl acetate, butyl acetate, dichloromethane, films, metal foils, vehicles, hard ketones (acetone, methyl ethyl ketone), butanol, ethanol, tetpapers, cardboard, pencils rahydrofuran, toluene, benzene, methyl acetate Cosmetics industry

Alcohols, esters

Rotogravure printing shops

Toluene, xylene, heptane, hexane

Textile printing

Hydrocarbons

Drycleaning shops

Hydrocarbons, tetrachloroethane

1636

21.1 Absorptive Solvent Recovery

Table 21.1.1. Application of organic solvents. [After references 2-4]. Application

Solvents

Sealing board manufacture

Hydrocarbons, toluene, trichloroethane

Powder and explosives production

Ether, alcohol, benzene

Degreasing and scouring applications Metal components Leather Wool Loading and reloading facilities (tank venting gases)

Hydrocarbons, trichloroethylene, tetrachloroethylene, dichloromethane

Tetrachloroethylene Tetrachloroethane Gasoline, benzene, toluene, xylene

The basic principles of some generally accepted gas cleaning processes for solvent removal are given in Table 21.1.2. For more than 70 years adsorption processes using activated carbon, in addition to absorption and condensation processes, were dominant in removal and recovery of solvents. The first solvent recovery plant for acetone was commissioned for economic reasons in 1917 by Bayer. In the decades that followed, solvent recovery plants were built and operated only if the value of the recovered solvents exceeded the operation costs and depreciation of the plant. Today such plants are used for adsorptive purification of exhaust air streams, even if the return is insufficient, to meet environmental and legal requirements. Table 21.1.2. Cleaning processes for the removal and recovering of solvents.

Basic principle Only waste heat recovery possible

Incineration

The solvents are completely destroyed by: thermal oxidation (>10 g/m3) catalytic oxidation (3-10 g/m3)

Condensation

High concentration of solvents (>50 g/m3) - Effective in reducing heavy emissions, are lowered by direct or indirect condensa- - good quality of the recovered solvents, tion with refrigerated condensers - difficult to achieve low discharge levels

Absorption Scrubbers using non-volatile organics as scrubbing medium for the solvent removal. Desorption of the spent scrubbing fluid Membrane permeation

Membrane permeation processes are suited for small volumes with high concentrations, e.g., to reduce hydrocarbon concentration from several hundred g/m3 down to 5-10 g/m3

Advantageous if: - scrubbing fluid can be reused directly without desorption of the absorbed solvent - scrubbing with the subcooled solvent itself possible To meet the environmental protection requirements a combination with other cleaning processes (e.g. activated carbon adsorber) is necessary

Biological treatment Biological decomposition by means of Sensitive to temperature and solvent bacteria and using bioscrubber or concentration biofilter

Klaus-Dirk Henning

1637

Table 21.1.2. Cleaning processes for the removal and recovering of solvents.

Basic principle Adsorption

Adsorption of solvents by passing the waste air through an adsorber filled with activated carbon. Solvent recovery by steam or hot gas desorption

By steam desorption the resulting desorbate has to be separated in a water and an organic phase using a gravity separator (or a rectification step)

This contribution deals with adsorptive solvent removal,2,12,13 recovery, and some other hybrid processes using activated carbon in combination with other purification steps. Due to more stringent environmental protection requirements, further increases in the number of adsorption plants for exhaust air purification may be expected. 21.1.2 BASIC PRINCIPLES 21.1.2.1 Fundamentals of adsorption To understand the adsorptive solvent recovery we have to consider some of the fundamentals of adsorption and desorption1-9 (Figure 21.1.1). Adsorption is the term for the enrichment of gaseous or dissolved substances on the boundary surface of a solid (the adsorbent). The surfaces of adsorbents have what we call active sites where the binding forces between the individual atoms of the solid structure are not completely saturated by neighboring atoms. These active sites can bind foreign molecules which, when Figure 21.1.1. Adsorption terms [After reference 2]. bound, are referred to as adsorpt. The overall interphase boundary is referred to as adsorbate. The term, desorption, designates the liberation of the adsorbed molecules, which is a reversal of the adsorption process. The desorption process generates desorbate which consists of the desorpt and the desorption medium.2 A distinction is made between two types of adsorption mechanisms: chemisorption and physisorption. Physisorption, is reversible and involves exclusively physical interaction forces (van der Waals forces) between the component to be adsorbed and the adsorbent. Chemisorption, is characterized by greater interaction energies which result in a chemical modification of the component adsorbed along with its reversible or irreversible adsorption.2 Apart from diffusion through the gas-side of boundary film, adsorption kinetics are predominantly determined by the diffusion rate through the pore structure. The diffusion coefficient is governed by both the ratio of pore diameter to the radius of the molecule being adsorbed and other characteristics of the component being adsorbed. The rate of adsorption is governed primarily by diffusional resistance. Pollutants must first diffuse from the bulk gas stream across the gas film to the exterior surface of the adsorbent (gas film resistance).

1638

21.1 Absorptive Solvent Recovery

Due to the highly porous nature of the adsorbent, its interior contains by far the majority of the free surface area. For this reason, the molecule to be adsorbed must diffuse into the pore (pore diffusion). After their diffusion into the pores, the molecules must then physically adhere to the surface, thus losing the bulk of their kinetic energy. Adsorption is an exothermic process. The adsorption enthalpy decreases as a load of adsorbed molecules increases. In activated carbon adsorption systems for solvent recovery, the liberated adsorption enthalpy normally amounts to 1.5 times the evaporation enthalpy at the standard working capacities which can result in a 20 K or more temperature increase. In the process, exothermic adsorption mechanisms may coincide with endothermic desorption mechanisms.2 21.1.2.2 Adsorption capacity The adsorption capacity (adsorptive power, loading) of an adsorbent resulting from the size and the structure of its inner surface area for a defined component is normally represented by a function of the component concentration c in the carrier gas for the equilibrium conditions at a constant temperature. This is known as the adsorption isotherm x=f(c)T. There are a variety of approaches proceeding from different model assumptions for the quantitative description of adsorption isotherms (Table 21.1.3) which are discussed in detail in the literature.2,9,11 Table 21.1.3. Overview of major isotherm equations [After reference 2]. Adsorption isotherm by

Isotherm equation n

Notes

Freundlich

x = k × pi

low partial pressures of the component to be adsorbed

Langmuir

x max bp i x = --------------------1 + bp i

homogeneous adsorbent surface, monolayer adsorption

Brunauer, Emmet, Teller (BET)

pi 1 C–1 p ------------------------ = ----------------- + ----------------- -----i V ( ps – pi ) V MB C V MB C p s

homogeneous surface, multilayer adsorption, capillary condensation

Dubinin-Radushkevich

n RT  p  log V = log V s – 2.3  ---------  log ----s-  ε β  pi o

potential theory; n = 1,2,3; e.g., n = 2 for the adsorption of organic solvent vapors on microporous activated carbon grades

where: K, n − specific constant of the component to be adsorbed; x − adsorbed mass, g/100 g; xmax − saturation value of the isotherm with monomolecular coverage, g/100 g; b − coefficient for component to be adsorbed, Pa; C − constant; T − temperature, K; pi − partial pressure of component to be adsorbed, Pa; ps − saturation vapor pressure of component to be adsorbed, Pa; V − adsorpt volume, ml; VMB − adsorpt volume with monomolecular coverage, ml; Vs − adsorpt volume on saturation, ml; VM - molar volume of component to be adsorbed, l/mol; R − gas constant, J(kg K); εo − characteristic adsorption energy, J/kg; β − affinity coefficient

The isotherm equation developed by Freundlich on the basis of empirical data can often be helpful. According to the Freundlich isotherm, the logarithm of the adsorbent loading increases linearly with the concentration of the component to be adsorbed in the carrier gas. This is also the result of the Langmuir isotherm which assumes that the bonding energy rises exponentially as loading decreases.

Klaus-Dirk Henning

1639

However, commercial adsorbents do not have a smooth surface but rather are highly porous solids with a very irregular and rugged inner surface. This fact is taken into account by the potential theory which forms the basis of the isotherm equation introduced by Dubinin. At adsorption temperatures below the critical temperature of the component to be adsorbed, the adsorbent pores may fill up with liquid adsorpt. This phenomenon is known as capillary condensation and enhances the adsorption capacity of the adsorbent. Assuming cylindrical pores, capillary condensation can be quantitatively described with the aid of the Kelvin equation, the degree of pore filling being inversely proportional to the pore radius. If several compounds are contained in a gas stream the substances compete for the adsorption area available. In that case, compounds with large interacting forces may displace less readily adsorbed compounds from the internal surface of the activated carbon.

Figure 21.1.2. The idealized breakthrough curve of a fixed bed adsorber.

As previously stated, the theories of adsorption are complex, with many empirically determined constants. For this reason, pilot data should always be obtained on the specific pollutant adsorbent combination prior to full-scale engineering design. 21.1.2.3 Dynamic adsorption in adsorber beds For commercial adsorption processes such as solvent recovery not only the equilibrium load, but also the rate at which it is achieved (adsorption kinetics) is of decisive importance.9-15 The adsorption kinetics is determined by the rates of the following series of individual steps:

1640

21.1 Absorptive Solvent Recovery

•

transfer of molecules to the external surface of the adsorbent (border layer film diffusion) • diffusion into the particle • adsorption The adsorbent is generally in a fixed adsorber bed and the contaminated air is passed through the adsorber bed (Figure 21.1.2). When the contaminated air first enters through the adsorber bed most of the adsorbate is initially adsorbed at the inlet of the bed and the air passes on with little further adsorption taking place. Later, when the inlet of the absorber becomes saturated, adsorption takes place deeper in the bed. After a short working time the activated carbon bed becomes divided into three zones: • the inlet zone where the equilibrium load corresponds to the inlet concentration, co • the adsorption or mass transfer zone, MTZ, where the adsorbate concentration in the air stream is reduced • the outlet zone where the adsorbent remains incompletely loaded In time the adsorption zone moves through the activated carbon bed.18-21 When the MTZ reaches the adsorber outlet zone, breakthrough occurs. The adsorption process must be stopped if a predetermined solvent concentration in the exhaust gas is exceeded. At this point the activated carbon bed has to be regenerated. If, instead the flow of contaminated air is continued on, the exit concentration continues to rise until it becomes the same as the inlet concentration. It is extremely important that the adsorber bed should be at least as long as the mass transfer zone of the component to be adsorbed. The following are key factors in dynamic adsorption and help determine the length and shape of the MTZ: • the type of adsorbent • the particle size of the adsorbent (may depend on maximum allowable pressure drop) • the depth of the adsorbent bed • the gas velocity • the temperature of the gas stream and the adsorbent • the concentration of the contaminants to be removed • the concentration of the contaminants not to be removed, including moisture • the pressure of the system • the removal efficiency required • possible decomposition or polymerization of contaminants on the adsorbent 21.1.2.4 Regeneration of the loaded adsorbents In the great majority of adsorptive waste gas cleaning processes, the adsorpt is desorbed after the breakthrough loading has been attained and the adsorbent reused for pollutant removal. Several aspects must be considered when establishing the conditions of regeneration for an adsorber system.2-4,6,13-16 Very often, the main factor is an economic one, to establish that an in-place regeneration is or is not preferred to the replacement of the entire adsorbent charge. Aside from this factor, it is important to determine if the recovery of the contaminant is worthwhile, or if only regeneration of the adsorbent is required. The pro-

Klaus-Dirk Henning

1641

cess steps required for this purpose normally dictate the overall concept of the adsorption unit. Regeneration, or desorption, is usually achieved by changing the conditions in the adsorber to bring about a lower equilibrium-loading capacity. This is done by either increasing the temperature or decreasing the partial pressure. For regeneration of spent activated carbon from waste air cleaning processes, the following regeneration methods are available (Table 21.1.4). Table 21.1.4. Activated carbon regeneration processes [After references 2,16]. Regeneration method Pressure-swing process

Principle

Application examples

Special features

Alternating between elevated pressure during adsorption and pressure reduction during desorption

Gas separation, e.g., N2 from the air, CH4 from biogas, CH4 from hydrogen

Raw gas compression; lean gas stream in addition to the product gas stream

Steam desorption or inert gas desorption at temperatures of < 500°C Partial gasification at 800 to 900°C with steam or other suitable oxidants

Solvent recovery, process Reprocessing of desorbate waste gas cleanup

Extraction with carbon disulfide or other solvents Percolation with caustic soda e.g., extraction with supercritical CO2

Sulfur extraction Sulfosorbon process Phenol-loads activated carbon Organic compounds

Temperature-swing process:

Desorption

Reactivation

Extraction with: Organic solvents Caustic soda solution Supercritical gases

All organic compounds Post-combustion, if adsorbed in gas cleaning required scrubbing of flue applications gas generated Desorbate treatment by distillation, steam desorption of solvent Phenol separation with subsequent purging Separation of CO2/ organic compounds

• Pressure-swing process • Temperature-swing process • Extraction process In the solvent recovery plants, temperature-swing processes are most frequently used. The loaded adsorbent is directly heated by the stream of hot inert gas, which, at the same time, serves as a transport medium to discharge the desorbed vapor and reduce the partial pressure of the gas-phase desorpt. As complete desorption of the adsorpt cannot be accomplished in a reasonable time in commercial-scale systems, there is always heel remaining which reduces the adsorbent working capacity. Oxidic adsorbents offer the advantage that their adsorptive power can be restored by controlled oxidation with air or oxygen at high temperatures. 21.1.3 COMMERCIALLY AVAILABLE ADSORBENTS Adsorbers used for air pollution control and solvent recovery predominantly employ activated carbon. Molecular sieve zeolites are also used. Polymeric adsorbents can be used but are seldom seen.

1642

21.1 Absorptive Solvent Recovery

21.1.3.1 Activated carbon Activated carbon1-8 is the trade name for a carbonaceous adsorbent which is defined as follows: Activated carbons5 are non-hazardous, processed, carbonaceous products, having a porous structure and a large internal surface area. These materials can adsorb a wide variety of substances, i.e., they are able to attract molecules to their internal surface, and are therefore called adsorbents. The volume of pores of the activated carbons is generally greater than 0.2 ml g-1. The internal surface area is generally greater than 400 m2g-1. The width of the pores ranges from 0.3 to several thousand nm. All activated carbons are characterized by their ramified pore system within which various mesopores (r = 1-25 nm), micropores (r = 0.4-1.0 nm) and submicropores (r < 0.4 nm) each of which branch off from macropores (r > 25 nm).

Figure 21.1.3. Schematic model of activated carbon. [Adapted, by permission, from K.-D. Henning, S. Schäfer, Impregnated activated carbon for environmental protection, Gas Separation & Purification, 1993, Vol. 7, No. 4, p. 235.]

Figure 21.1.3 shows schematically the pore system important for adsorption and desorption.17 The large specific surfaces are created predominantly by the micropores. Activated carbon is commercially available in shaped (cylindrical pellets), granular, or powdered form. Due to its predominantly hydrophobic surface properties, activated carbon preferentially adsorbs organic substances and other non-polar compounds from gas and liquid phases. Activated alumina, silica gel, and molecular sieves will adsorb water preferentially from a gas-phase mixture of water vapor and an organic contaminant. In Europe, cylindrically-shaped activated carbon pellets with a diameter of 3 or 4 mm are used for solvent recovery because they assure a low-pressure drop across the adsorber system. Physical and chemical properties of three typical activated carbon types used in solvent recovery are given in Table 21.1.5.

Klaus-Dirk Henning

1643

Table 21.1.5 Activated carbon types for waste air cleaning [After reference 2]. Compacted density, kg/m3

Pore volume for pore size d < 20 nm, ml/g

Pore volume for pore size d > 20 nm, ml/g

Specific surface area, m2/g

Specific heat capacity, J/kgK

Intake air and exhaust air cleanup, odor control, adsorption of low-boiling hydrocarbons

400 - 500

0.5 - 0.7

0.3 - 0.5

1000-1200

850

Activated Solvent recovery, adsorpcarbon, tion of medium-high boilmedium-pore ing hydrocarbons

350 - 450

0.4 - 0.6

0.5 - 0.7

1200-1400

850

Activated carbon, wide-pore

300 - 400

0.3 - 0.5

0.5 - 1.1

1000-1500

850

Adsorbent Activated carbon, fine-pore

Applications (typical examples)

Adsorption and recovery of high-boiling hydrocarbons

21.1.3.2 Molecular sieve zeolites Molecular sieve zeolites2,10,11 are hydrated, crystalline aluminosilicates which give off their crystal water without changing their crystal structure so that the original water sites are free for the adsorption of other compounds. Activation of zeolites is a dehydration process accomplished by the application of heat in a high vacuum. Some zeolite crystals show behavior opposite to that of activated carbon in that they selectively adsorb water in the presence of nonpolar solvents. Zeolites can be made to have specific pore sizes that impose limits on the size and orientation of molecules that can be adsorbed. Molecules above a specific size cannot enter the pores and therefore cannot be adsorbed (steric separation effect). Substitution of the greater part of the Al2O3 by SiO2 yields zeolites suitable for the selective adsorption of organic compounds from high-moisture waste gases. 21.1.3.3 Polymeric adsorbents The spectrum of commercial adsorbents2 for use in air pollution control also includes beaded, hydrophobic adsorber resins consisting of nonpolar, macroporous polymers produced for the specific application by polymerizing styrene in the presence of a crosslinking agent. Crosslinked styrene-divinylbenzene resins are available on the market under different trade names. Their structure-inherent fast kinetics offers the advantage of relatively low desorption temperatures. They are insoluble in water, acids, lye, and a large number of organic solvents. 21.1.4 ADSORPTIVE SOLVENT RECOVERY SYSTEMS Stricter regulations regarding air pollution control, water pollution control, and waste management have forced companies to remove volatile organics from atmospheric emissions and workplace environments. But apart from compliance with these requirements, economic factors are decisive in solvent recovery. Reuse of solvent in production not only reduces operating cost drastically but may even allow profitable operation of a recovery system.

1644

21.1 Absorptive Solvent Recovery

21.1.4.1 Basic arrangement of adsorptive solvent recovery with steam desorption Adsorptive solvent recovery units have at least two, but usually, three or four parallel-connected fix-bed adsorbers which pass successively through the four stages of the operation cycle.1-4,18-23 • Adsorption • Desorption • Drying • Cooling Whilst adsorption takes place in one or more of them, desorption, drying, and cooling take place in the others. The most common adsorbent is activated carbon in the shape of 3 or 4 mm pellets or as granular type with a particle size of 2 to 5 mm (4 x 10 mesh). A schematic flow sheet of a two adsorber system for the removal of water-insoluble solvents is shown in Fig 21.1.4. The adsorber feed is pretreated if necessary to remove solids (dust), liquids (droplets or aerosols) or high-boiling components as these can hamper performance. Frequently, the exhaust air requires cooling. To prevent an excessive temperature increase across the bed Figure 21.1.4. Flow sheet of a solvent recovery system with steam due to the heat of adsorption, inlet desorption. solvent concentrations are usually limited to about 50 g/m3. In most systems, the solvent-laden air stream is directed upwards through a fixed carbon bed. As soon as the maximum permissible breakthrough concentration is attained in the discharge clean air stream, the loaded adsorber is switched to regeneration. To reverse the adsorption of the solvent, the equilibrium must be reversed by increasing the temperature and decreasing the solvent concentration by purging. For solvent desorption direct steaming of the activated carbon bed is the most widely used regeneration technique because it is cheap and simple. Steam (110-130°C) is very effective in raising the bed temperature quickly and is easily condensed to recover the solvent as a liquid. A certain flow is also required to reduce the partial pressure of the adsorbate and carry the solvent out of the activated carbon bed. A flow diagram for steam desorption for toluene recovery is given in Figure 21.1.5.

Klaus-Dirk Henning

1645

Figure 21.1.5. Principle of steam desorption.

Figure 21.1.6. Desorption efficiency and steam consumption.

First, the temperature of the activated carbon is increased to approx. 100°C. This temperature increase reduces the equilibrium load of the activated carbon. Further reduction of the residual load is obtained by the flushing effect of the steam and the declining toluene partial pressure. The load difference between spent and regenerated activated carbon − the “working capacity” − is then available for the next adsorption cycle. The counter-current pattern of adsorption and desorption favors high removal efficiencies. Desorption of the adsorbed solvents starts after the delay required to heat the activated carbon bed. The specific steam consumption increases as the residual load of the activated carbon decreases (Figure 21.1.6). For cost reasons, desorption is not run to completion. The desorption time is optimized to obtain the acceptable residual load with a minimum specific steam consumption. The amount of steam required depends on the

1646

21.1 Absorptive Solvent Recovery

interaction forces between the solvent and the activated carbon. The mixture of steam and solvent vapor from the adsorber is condensed in a condenser. If the solvent is immiscible with water the condensate is led to a gravity separator (making use of the density differential) where it is separated into an aqueous and solvent fraction. If the solvents are partly or completely miscible with water, additional special processes are necessary for solvent separation and treatment (Figure 21.1.7). For partly miscible solvents the desorpt steam/solvent mixture is separated by enrichment condensation, rectification or by a membrane processes. The separation of completely water-soluble solvents requires a rectification column.2

Figure 21.1.7. Options for solvent recovery by steam desorption [After references 2, 3,12].

After steam regeneration, the hot and wet carbon bed will not remove organics from air effectively, because high temperature and humidity do not favor complete adsorption. The activated carbon is therefore dried by a hot air stream. Before starting the next adsorption cycle the activated carbon bed is cooled to ambient temperature with air. Typical operating data for solvent recovery plants and design ranges are given in Table 21.1.6. Table 21.1.6. Typical operating data for solvent recovery plants. Operating data Air velocity (m/s): 0.2-0.4 Air temperature (°C): 20-40 Bed height (m): 0.8-2 Steam velocity (m/s): 0.1-0.2 Time cycle for each adsorber Adsorption (h): 2-6 Steaming (h): 0.5-1 Drying (hot air, h): 0.2-0.5 Cooling (cold air, h): 0.2-0.5

Range of design Solvent concentration (g/m3): 1-20 Solvent adsorbed by activated carbon per cycle (% weight): 10-20 Steam/solvent ratio: 2-5:1 Energy (kWh/t solvent): 50-600 Cooling water (m3/t solvent): 30-100 Activated carbon (kg/t solvent): 0.5-1

Klaus-Dirk Henning

1647

21.1.4.2 Designing solvent recovery systems 21.1.4.2.1 Design basis1-3,18-23 Designing an activated carbon system should be based on pilot data for the particular system and information obtained from the activated carbon producer. The basic engineering and the erection of the system should be done by an experienced engineering company. Solvent recovery systems would also necessitate the specification of condenser duties, distillation tower sizes, holding tanks and piping and valves. Designing of a solvent recovery system requires the following information (Table 21.1.7). Table 21.1.7. The design basis for adsorption systems by the example of solvent recovery [After references 1-3]. Site-specific conditions

Emission sources: Production process(es), continuous/shift operation Suction system: Fully enclosed system, negative pressure control Available drawings: Building plan, machine arrangement, available space, neighborhood Operating permit as per BimSchG (German Federal Immission Control Act)

Exhaust air data (max., min., normal

Volumetric flow rate (m3h) Relative humidity (%) Solvent concentration by components (g/m3) The concentration of other pollutants and condition (fibers, dust) Temperature (°C) Pressure (mbar)

Characterization of solvents

Molecular weight, density Critical temperature Water solubility Stability (chemical, thermal) Safety instructions as per German Hazardous Substances Regulations (description of hazards, safety instructions, safety data sheets) Physiological effects: irritating, allergenic, toxic, cancerogenic Boiling point, boiling range Explosion limits Corrosiveness Maximum exposure limits (MEL)

Utilities (type and availability)

Electric power Steam Cooling water Cooling brine Compressed air, instrument air Inert gas

Waste streams (avoidance, recycling, disposal)

Solvent Steam condensate Adsorbent Filter dust

Guaranteed performance data at rated load

Emission levels: pollutants, sound pressure level (dB(A)) Utility consumption, materials- Availability

1648

21.1 Absorptive Solvent Recovery

After this information has been obtained, the cycle time of the system must be determined. The overall size of the unit is determined primarily by economic considerations, balancing low operating costs against the capital costs (smaller system → shorter cycle time; bigger adsorber → long cycle time). The limiting factor for the adsorber diameter is normally the maximum allowable flow velocity under continuous operating conditions. The latter is related to the free vessel cross-section and, taking into account fluctuations in the gas throughput, selected such (0.1 to 0.5 m/s) that it is sufficiently below the disengagement point, i.e., the point at which the adsorbent particles in the upper bed area are set into motion. Once the adsorber diameter has been selected, the adsorber height can be determined from the required adsorbent volume plus the support tray and the free flow cross-sections to be provided above and below the adsorbent bed. The required adsorbent volume is dictated by the pollutant concentration to be removed per hour and the achievable adsorbent working capacity. From the view point of adsorption theory, it is important that the bed be deeper than the length of the transfer zone which is unsaturated. Any multiplication of the minimum bed depth gives more than proportional increase in capacity. In general, it is advantageous to size the adsorbent bed to the maximum length allowed by pressure drop considerations. Usually, fixed-bed adsorber has a bed thickness of 0.8-2 m. 21.1.4.2.2 Adsorber types Adsorbers for solvent recovery1-3 units (Figure 21.1.8) are built in several configurations: • the vertical cylindrical fixed-bed adsorber • the horizontal cylindrical fixed-bed adsorber • vertical annular bed adsorber • moving-bed adsorber • rotor adsorber (axial or radial) For the cleanup of high volume waste gas streams, multi-adsorber systems are preferred. Vertical adsorbers are most common. Horizontal adsorbers, which require less overhead space, are used mainly for large units. The activated carbon is supported on a cast-iron grid which is sometimes covered by a mesh. Beneath the bed, a 100 mm layer of 5-50mm size gravel (or Al2O3 pellets) is placed which acts as a heat accumulator and it stores part of the desorption energy input for the subsequent drying step. The depth of the carbon bed, chosen in accordance with the process conditions, is from 0.8 to 2 m. Adsorbers with an annular bed of activated carbon are less common. The carbon in the adsorber is contained between vertical, cylindrical screens. The solvent laden air passes from the outer annulus through the outer screen, the carbon bed and the inner screen then exhausts to atmosphere through the inner annulus. The steam for regeneration of the carbon flows in the opposite direction to the airstream. Figure 21.1.8 illustrates the operating principle of a counter-current flow vertical moving bed adsorber. The waste gas is routed in countercurrent to the moving adsorbent and such adsorption systems have a large number of single moving beds. The discharge system ensures a uniform discharge (mass flow) of the loaded adsorbent. In this moving bed system adsorption and desorption are performed in a separate column or column sections. The process relies on an abrasion resistant adsorbent with good flow properties (beaded carbon or adsorber resins). After the adsorption stage

Klaus-Dirk Henning

1649

Figure 21.1.8. Adsorber construction forms.

of a multi-stage fluidized-bed system, the loaded adsorbent is regenerated by means of steam or hot inert gas in a moving-bed regenerator and pneumatically conveyed back to the adsorption section.2 When using polymeric adsorbents in a comparable system, desorption can be carried out with heated air because the structure-inherent faster kinetics of these adsorbents require desorption temperatures of only 80 to 100°C.

1650

21.1 Absorptive Solvent Recovery

Rotor adsorbers allow simultaneous adsorption and desorption in different zones of the same vessel without the need for adsorbent transport and the associated mechanical stresses. Rotor adsorbers can be arranged vertically or horizontally and can be designed for axial or radial gas flow so that they offer great flexibility in the design of the adsorption system. The choice of the rotor design2,27,33 depends on the type of adsorbent employed. Rotor adsorbers accommodating pellets or granular carbon beds come as wheel type baskets divided into segments to separate the adsorption zone from the regeneration zone. Alternatively, they may be equipped with a structured carrier (e.g., ceramics, cellulose) coated with or incorporating different adsorbents such as • activated carbon fibers • beaded activated carbon • activated carbon particles • particles of hydrophobic zeolite The faster changeover from adsorption to desorption compared with fixed-bed adsorbers normally translates into a smaller bed volume and hence, a reduced pressure loss. Hot air desorption is also the preferred method for rotor adsorbers. Hot air rates are about one-tenth of the exhaust air rate. The desorbed solvents can either be condensed by chilling or disposed of by high-temperature or catalytic combustion. In some cases, removal of the concentrated solvents from the desorption air in a conventional adsorption system and subsequent recovery for reuse may be a viable option. The structural design of the adsorber is normally governed by the operating conditions during the desorption cycle when elevated temperatures may cause pressure loads on the materials. Some solvents when adsorbed on activated carbon, or in the presence of steam, oxidize or hydrolyze to a small extent and produce small amounts of corrosive materials. As a general rule, hydrocarbons and alcohols are unaffected and carbon steel can be used. In the case of ketones and esters, stainless steel or other corrosion resistant materials are required for the solvent-wetted parts of the adsorbers, pipework, and condensers, etc. For carbon tetrachloride and 1,1,1-trichloroethane and other chlorinated solvents, titanium has a high degree of resistance to the traces of acidity formed during recovery. 21.1.4.2.3 Regeneration The key to the effectiveness of solvent recovery is the residual adsorption capacity (working capacity) after in-place regeneration. The basic principles of regeneration are given in Sections 21.1.2.4 and 21.1.4.1. After the adsorption step has concentrated the solvent, the design of the regeneration system1-3,13-16 depends on the application and the downstream use of the desorbed solvents whether it be collection and reuse or destruction. Table 21.1.8 may give some indications for choosing a suitable regeneration step.

Klaus-Dirk Henning

1651

Table 21.1.8. The basic principle of regeneration. Regeneration step

Basic principle

Typical applications

Off-site reactivation

Transportation of the spent carbon to a reactivation plant. Partial gasification at 800-900°C with steam in a reactivation reactor. Reuse of the regenerated activated carbon

Low inlet concentration odor control application solvents with boiling points above 200°C polymerization of solvents

Steam desorption

Increasing the bed temperature by direct steaming and desorption of solvent. The steam is condensed and solvent recovered.

Most widely used regeneration technique for solvent recovery plants. Concentration range from 1-50 g/m3

Hot air desorption

Rotor adsorber are often desorbed by hot air

Concentration by adsorption and incineration of the high calorific value desorbate

Hot inert gas desorption

Desorption by means of a hot inert gas recirculation. Solvent recovery can be achieved with refrigerated condenser

Recovery of partly or completely water soluble solvents. Recovery of reactive solvents (ketones) or solvents which react with hot steam (chlorinated hydrocarbons)

Pressure swing Atmospheric pressure adsorption and heatless vacuum regeneration. The revaporized adsorption/ vacuum regeneration and desorbed vapor is transported from the adsorption bed via a vacuum pump

Low-boiling and medium-boiling hydrocarbon vapor mixes from highly concentrated tank venting or displacement gases in large tank farms. Recovery of monomers in the polymer industry

Reduced pressure Increasing the bed temperature by direct and low-temperature steaming and desorption of solvent. The steam desorption steam is condensed and solvent recovered.

Recovery of reactive solvents of the ketone-type. Minimizing of side-reactions like oxidizing, decomposing or polymerizing.

21.1.4.2.4 Safety requirements Adsorption systems including all the auxiliary and ancillary equipment needed for their operation are subject to the applicable safety legislation and the safety requirements of the trade associations and have to be operated in accordance with the respective regulations.2,19 Moreover, the manufacturer's instructions have to be observed to minimize safety risks. Activated carbon or its impurities catalyze the decomposition of some organics, such as ketones, aldehydes, and esters. These exothermic decomposition reactions25,26 can generate localized hot spots and/or bed fires within an adsorber if the heat is allowed to build up. These hazards will crop up if the flow is low and the inlet concentrations are high, or if an adsorber is left dormant without being completely regenerated. To reduce the hazard of bed fires, the following procedures are usually recommended: • Adsorption of readily oxidizing solvents (e.g., cyclohexanone) require increased safety precautions. Instrumentation, including alarms, should be installed to monitor the temperature change across the adsorber bed and the outlet CO/CO2 concentrations. The instrumentation should signal the first signs of decomposition so that any acceleration leading to a bed fire can be forestalled. Design parameters should be set so as to avoid high inlet concentrations and low flow. A

1652

21.1 Absorptive Solvent Recovery

• •

• • • • • • •

minimum gas velocity of approximately 0.2 m/s should be maintained in the fixed-bed adsorber at all times to ensure proper heat dissipation. A virgin bed should be steamed before the first adsorption cycle. Residual condensate will remove heat. Loaded activated carbon beds require constant observation because of the risk of hot spot formation. The bed should never be left dormant unless it has been thoroughly regenerated. After each desorption cycle, the activated carbon should be properly cooled before starting a new adsorption cycle. Accumulation of carbon fines should be avoided due to the risk of local bed plugging which may lead to heat build-up even with weak exothermic reactions. CO and temperature monitors with alarm function should be provided at the clean gas outlet. The design should also minimize the possibility of explosion hazard Measurement of the internal adsorber pressure during the desorption cycle is useful for monitoring the valve positions. Measurement of the pressure loss across the adsorber provides information on particle abrasion and blockages in the support tray. Adsorption systems must be electrically grounded.

21.1.4.3 Special process conditions 21.1.4.3.1 Selection of the adsorbent The adsorption of solvents on activated carbon1-4 is controlled by the properties of both the carbon and the solvent and the contacting conditions. Generally, the following factors, which characterize the solvent-containing waste air stream, are to be considered when selecting the most well-suited activated carbon quality for waste air cleaning: for solvent/waste air: • solvent type (aliphatic/aromatic/polar solvents) • concentration, the partial pressure • molecular weight • density • boiling point, boiling range • critical temperature • desorbability • explosion limits • thermal and chemical stability • water solubility • adsorption temperature • adsorption pressure • solvent mixture composition • solvent concentration by components • humidity • impurities (e.g., dust) in the gas stream

Klaus-Dirk Henning

1653

for the activated carbon type: General properties: • apparent density • particle size distribution • hardness • surface area • activity for CCl4/benzene • the pore volume distribution curve For the selected solvent recovery task: • the form of the adsorption isotherm • the working capacity • and the steam consumption The surface area and the pore size distribution are factors of primary importance in the adsorption process. In general, the greater the surface area, the higher the adsorption capacity will be. However, that surface area within the activated carbon must be accessible. At low concentration (small molecules), the surface area in the smallest pores, into which the solvent can enter, is the most efficient surface. With higher concentrations (larger molecules) the larger pores become more efficient. At higher concentrations, capillary condensation will take place within the pores and the total micropore volume will become the limiting factor. These molecules are retained at the surface in the liquid state, because of intermolecular or van der Waals forces. Figure 21.1.9 shows the relationship between maximum effective pore size and concentration for the adsorption of toluene according to the Kelvin theory.

Figure 21.1.9. The relationship between maximum effective pore diameter and toluene concentration.

It is evident that the most valuable information concerning the adsorption capacity of a given activated carbon is its adsorption isotherm for the solvent being adsorbed and its

1654

21.1 Absorptive Solvent Recovery

pore volume distribution curve. Figure 21.1.10 presents idealized toluene adsorption isotherms for three carbon types: • large pores predominant • medium pores predominant • small pores predominant

Figure 21.1.10. Idealized toluene adsorption isotherms.

The adsorption lines intersect at different concentrations, depending on the pore size distribution of the carbon. The following applies to all types of activated carbon: As the toluene concentration in the exhaust air increases, the activated carbon load increases. From the viewpoint of adsorption technology, high toluene concentration in the exhaust air is more economic than a low concentration. But to ensure safe operation, the toluene concentration should not exceed 40% of the lower explosive limit. There is, however, an optimum activated carbon type for each toluene concentration (Figure 21.1.10). The pore structure of the activated carbon must be matched to the solvent and the solvent concentration for each waste air cleaning problem. Figure 21.1.11 shows Freundlich adsorption isotherms of five typical solvents as a function of the solvent concentrations in the gas stream. It is clear that different solvents are adsorbed at different rates according to the intensity of the interacting forces between solvent and activated carbon.

Klaus-Dirk Henning

1655

Figure 21.1.11. Adsorption isotherm of five typical solvents (After reference 13).

Physical-chemical properties of three typical activated carbons used in solvent recovery appear in Table 21.1.9. Table 21.1.9. Activated carbon pellets for solvent recovery [After reference 13]. AC-Type

C 38/4

Appearance

C 40/4

D 43/4

cylindrically shaped

Bulk density (shaken), kg/m

3

380±20

400±20

430±20

Moisture content, wt%

<5.0

<5.0

<5.0

Ash content, wt%

<6.0

<6.0

<5.0

4

4

4

Surface area (BET), m /g

>1250

>1250

>1100

Carbon tetrachloride activity, wt%

80±3

75±3

67±3

Particle diameter, mm 2

Producers13 of activated carbon are in the best position to provide technical advice on • selecting the right activated carbon type • contributing to the technical and economic success of a solvent recovery plant 21.1.4.3.2 Air velocity and pressure drop In solvent recovery systems, air velocity rates through the bed should be between 0.2 to 0.4 m/s. The length of the MTZ is directly proportional to the air velocity. Lower veloci-

1656

21.1 Absorptive Solvent Recovery

ties (<0.2 m/s) would lead to better utilization of the adsorption capacity of the carbon, but there is a danger that the heat of adsorption is not to be carried away, which would cause overheating and possibly ignition of the carbon bed. The power to operate the blowers to move waste air through the system constitutes one of the major operating expenses of the system. Pressure drop (resistance to flow) across the system is a function of • air velocity • bed depth • activated carbon particle size Small particle size activated carbon will produce a high-pressure drop through the activated-carbon bed. Figure 21.1.12 compares the pressure drop of cylindrically-shaped activated carbon pellets with activated-carbon granulates. The activated carbon particle diameter must not be excessively large because the long diffusion distances would delay adsorption and desorption. Commercially, cylindrical pellets with a particle diameter of 3 to 4 mm have been most efficient.

Figure 21.1.12. Pressure drop for various activated carbon types.

21.1.4.3.3 Effects of solvent-concentration, adsorption temperature and pressure For safety reasons, the concentration of combustible solvent vapors should be less than 50% of the lower explosive limit. The adsorption capacity of adsorbents increases as the concentration of the solvents increases. But the length of the MTZ is proportional to the solvent concentration. Because of the adsorption heat, as the adsorption front moves through the bed also a temperature front follows in the same direction. To deal with the adsorption heat the inlet solvent concentration is usually limited to about 50 g/m3.

Klaus-Dirk Henning

1657

The adsorption capacity of the adsorbent increases with pressure because the partial pressure of the solvent increases. An increase in adsorber temperature causes a reduction in adsorption capacity. Because the equilibrium capacity is lower at higher temperatures, the dynamic capacity (working capacity) of the activated carbon adsorber will also be lower. To enhance adsorption, the inlet temperature of the adsorber should be in the range of 20-40°C. In Figure 21.1.13 the adsorption isotherms of tetrahydrofuran on activated carbon D43/3 for several temperatures are shown.13

Figure 21.1.13. Adsorption isotherm of tetrahydrofuran for several temperatures.

21.1.4.3.4 Influence of humidity Activated carbon is basically hydrophobic; it adsorbs preferably organic solvents. The water adsorption isotherm (Figure 21.1.14) reflects its hydrophobic character. Below a relative gas humidity of about 40% co-adsorption of water can be neglected in most applications. However, higher humidity of the waste gases may affect the adsorption capacity of the activated carbon. Figure 21.1.15 demonstrates, using toluene as an example of how the relative humidity of the exhaust air influences the activated carbon loads. Experimental work24 has shown that • relative humidity rates below 30% will neither reduce the adsorption capacity nor the adsorption time, •. relative humidity rates above 70% substantially reduce the adsorption capacity, •. the humidity of the gas will affect the adsorption capacity much more with low toluene concentrations than with high concentrations, •. with toluene concentrations between 10 and 20 g/m3 the negative influence of the humidity is small.

1658

21.1 Absorptive Solvent Recovery

Figure 21.1.14. Water isotherm on activated carbon D43/4.

Figure 21.1.15. Dynamic toluene adsorption from humidified air [after reference 24].

The water content of the activated carbon after desorption may constitute another problem. The purification efficiency of each activated carbon is the better the less water is present after desorption. Unfortunately, the desorbed activated carbon in the vicinity of the adsorber walls usually contains high proportions of water (approx. 10 to 20%). With such a high water content, it is difficult to remove all the water even when drying with hot-air over longer periods. The wet and poorly-regenerated activated carbons in these zones fre-

Klaus-Dirk Henning

1659

quently lead to higher solvent concentrations in the purified air, and this is even at the beginning of the adsorption cycle. Some proposals for improvement include: • good insulation of the adsorber walls, • sufficient drying after each regeneration cycle, • cycle the desorption steam and the drying air counter-current to the direction of the adsorptive stream. 21.1.4.3.5 Interaction between solvents and activated carbon The majority of solvents are effectively recovered by adsorption on activated carbon and, when this is the case, the operation of the plant is straightforward. But some solvents may decompose, react or polymerize when in contact with activated carbon during the adsorption step and the subsequent steam desorption.19,25,26 Chlorinated hydrocarbon solvents can undergo hydrolysis to varying degrees on carbon surfaces, resulting in the formation of hydrogen chloride. Each activated carbon particle which is in contact with a metal screen or other constructional component may then act as a potential galvanic cell in the presence of moisture and chloride ions. Carbon disulfide can be catalytically oxidized to sulfur which remains on the internal surface of the activated carbon. Esters such as ethyl acetate are particularly corrosive because their hydrolysis results in the formation of acetic acid (or other acids). Aldehydes and phenol or styrene will undergo some degree of polymerization when in contact with hot activated carbon. During the adsorptive removal of ketones such as methyl ethyl ketone or cyclohexanone a reduced adsorption capacity has been measured. Corrosion problems were also apparent and in some cases, even spontaneous ignition of the activated carbon occurred.

Figure 21.1.16. Surface reaction of cyclohexanone on activated carbon [After reference 25].

1660

21.1 Absorptive Solvent Recovery

Activated carbon acts, in the presence of oxygen, as a catalyst during adsorption and even more frequently during desorption because of higher temperature.25,26 Figure 21.1.16 shows that the catalytic reactions occurring with Figure 21.1.17. Surface reaction of methyl ethyl ketone on activated cyclohexanone are predominantly carbon [After reference 25]. of the oxidation type. Adipic acid is first formed from cyclohexanone. Adipic acid has a high boiling point (213°C at a 13 mbar vacuum). Further products identified were: cyclopentanone as a degradation product from adipic acid, phenol, toluene, dibenzofuran, aliphatic hydrocarbons and carbon dioxide. The high-boiling adipic acid cannot be desorbed from the activated carbon in a usual steam desorption. Consequently, the life of the activated carbon becomes further reduced after each adsorption/ desorption cycle. During adsorption of methyl ethyl ketone in the presence of oxygen, some catalytic surface reactions occur (Figure 21.1.17). The reaction products are acetic acid, diacetyl, and presumably also diacetyl peroxide. Diacetyl has an intensive green coloration and a distinctive odor. Important for the heat balance of the adsorber is that all of the reactions are strongly exothermic. Unlike the side products of cyclohexanone, acetic acid the reaction product of the MEK (corrosive!) is removed from the activated carbon during steam desorption. Adsorption performance is thus maintained even after a number of cycles. Operators of ketones recovery plants should adhere to certain rules and also take some precautions.25 • Adsorption temperature should not be higher than 30°C (the surface reaction rates increase exponentially with temperature). • The above requirement of low-adsorption temperatures includes: a adequate cooling after the desorption step; b flow velocity of at least 0.2 m/s in order to remove the heat of adsorption. Since ketones will, even in an adsorbed state, oxidize on the activated carbon due to the presence of oxygen, these further precautions should be taken: • keep the adsorption/desorption cycles as short as possible, • desorb the loaded activated carbon immediately at the lowest possible temperatures, • desorb, cool and blanket the equipment with inert gas before an extended shutdown. 21.1.4.3.6 Activated carbon service life Activated carbon in solvent recovery service will have a useful service life of 1 to 10 years, depending on the attrition rate and reduction in adsorption capacity. Attrition rates are usually less than 1-3% per year. The actual rate will depend on carbon hardness. Particle abrasion and the resulting bed compaction leads to an increased

Klaus-Dirk Henning

1661

pressure loss after several years of service. After 3-5 years of service screening to its original size is necessary. Adsorption capacity can be reduced by traces of certain high-boiling materials (resins, volatile organosilicon compounds) in the waste air which are not removed during the desorption cycle. Some solvents may decompose, react or polymerize when in contact with activated carbon and steam (Section 21.1.4.3.5). This gradual loading will decrease the carbon’s activity. While carbon can remain in service in a reduced capacity state, it represents a non-optimal operation of the system that results in • reduced amount of solvent removed per cycle • increased steam consumption • higher emissions • shorter adsorption cycles Eventually, recovering capacity will diminish; it is more economical to replace carbon than continue its use in a deteriorated state. In replacing spent activated carbon there are two options: • replacing with virgin carbon and disposal of the spent carbon. • off-site reactivation of the spent carbon to about 95% of its virgin activity for about one-half of the cost of a new carbon Some activated carbon producers offer a complete service including carbon testing, off-site reactivation, transportation, adsorber-filling and carbon make-up.13 21.1.5 EXAMPLES FROM DIFFERENT INDUSTRIES Some examples of solvent recovery systems in different industries show that reliable and trouble-free systems are available. 21.1.5.1 Rotogravure printing shops Process principle Adsorptive solvent recovery with steam desorption and condensation units with gravity separator and a stripper have become a standard practice in modern production plants. The solvents-laden air (toluene, xylene) is collected from emission points, e.g., rotogravure printing machines, drying ducts by means of a blower and passed through the recovery plant. Design example30 In 1995, one of Europe's most modern printing shops in Dresden was equipped with an adsorptive solvent recovery system (Supersorbon® process (Figure 21.1.18.)). Design data (First stage of completion) Exhaust air flow rate 240,000 m3/h (expandable to 400,000 m3/h) Solvent capacity 2,400 kg/h (toluene) Exhaust air temperature 40°C max. 50 mg/m3 Toluene concentration in clean air (half-hour mean)

1662

21.1 Absorptive Solvent Recovery

Figure 21.1.18. Supersorbon® process for toluene recovery [After reference 30].

Solvent-laden air is exhausted at the three rotogravure printing presses by several fans operating in parallel and is routed in an upward flow through four adsorbers packed with Supersorbon® activated carbon. The solvent contained in the air is adsorbed on the activated carbon bed. Adsorption continues until breakthrough when the full retentive capacity of the adsorbent for solvent vapors is used up. The purity of the clean exhaust air complies with the regulation for emission level of less than 50 mg/m3. Regeneration of the adsorbent takes place by desorbing the solvent with a countercurrent flow of steam. The mixture of water and toluene vapors is condensed and the toluene, being almost insoluble in water, is separated by reason of its different density in a gravity separator. The recovered toluene can be reused in the printing presses without further treatment. Very small amounts of toluene dissolved in the wastewater are removed by stripping with air and returned to the adsorption system inlet together with the stripping air flow. The purified condensate from the steam is employed as make-up water in the cooling towers. Operating experience The adsorption system is distinguished by an extremely economical operation with a good ratio of energy consumption to toluene recovery. Solvent recovery rate is about 99.5%. 21.1.5.2 Packaging printing industry In the packaging printing industry solvents like ethyl acetate, ethanol, ketones, tetrahydrofuran (THF), hexane and toluene are used in printing. The solvent laden air generally contains between 2 and 15 g/m3 of solvent and, depending on the season, 5-18 g/m3 of water. Adsorptive recovery with steam desorption was widely used for many years but its disadvantage is that the solvents are recovered in various mixtures containing large amounts of water. The recovery of solvents is complicated because most solvents produce azeotropes with water which are not easily separated and, consequently, the wastewater fails to meet the more stringent environmental regulations. In the last 10 years, the adsor-

Klaus-Dirk Henning

1663

bent regeneration with hot inert gas has become the more favored process in the packaging printing industry. 21.1.5.2.1 Fixed bed adsorption with circulating hot gas desorption31,32 Activated carbon preferentially adsorbs non-polar organic solvents, while adsorbing relatively little water. Thus activated carbon allows the passage of 95-98% of the water while retaining the solvents from the gas stream. During the adsorption step solvent laden air or gas is delivered to the bed through the appropriate valves. As adsorption starts on a fresh bed, the effluent gas contains traces of solvent. Through time, the solvent level increases and when it reaches a predetermined value, the adsorption is stopped by closing the feed gas valves. The adsorption step usually lasts for 4-16 hours. An example of a solvent recovery plant with 5 activated carbon beds and hot gas desorption is shown in Figure 21.1.19.

Figure 21.1.19. Fixed bed adsorption with circulating hot gas desorption [After references 31,32].

If necessary, feed air is cooled to 35°C and sent by the blower V-1 to the main heater 6 from which it enters 4 beds simultaneously, while one bed is being regenerated. The cleaned air is collected in header 7 and vented. After the adsorption step, the air is displaced from the bed by an inert gas such as nitrogen. A circulating inert gas at 120-240°C serves for heating and stripping the solvent from the carbon. The solvent is recovered by cooling and chilling the circulating gas. The optimum chilling temperature depends on the boiling point of the solvent. Generally, chilling temperatures are between +10 and −30°C. During the heating step, the temperature of the effluent gas gradually increases until a predetermined value (for example, 150°C) is reached, after which the heater is bypassed and the bed is cooled down by cold gas. Typical solvent recovery plants will have from 2

1664

21.1 Absorptive Solvent Recovery

to 8 beds. All beds go through the same adsorption, inertization, heating and cooling steps but each at a different time. Referring to the flow sheet in Figure 21.1.19, the regeneration gas is circulated by blower V-2 to the gas heater, the header 8 to the bed being regenerated. The effluent gas is passed to the header 9, cooler, molecular sieves (or water condenser), to the chiller for solvent condensation and back to the blower V-2. At the end of the adsorption step, the bed contains 10-30% solvent and 1-2% water. There are various ways to remove the water separately and to recover a solvent containing between 0.1 and 1.5% water (Table 21.1.10). Table 21.1.10. Process conditions and water content in the solvent [After references 32,33]. Water in the Investment raw solvent cost, %

Heat needed, kWh/kg S

Power, kWh/kg S

Basic process with molsieve beds

0.1

100

2.2

1.0

Chiller instead of molsieve

1.5

91

1.5

1.0

Chiller and 1 carbon bed Chiller, and 1 carbon bed molsieve

0.4-1.0

93

1.1

1.0

0.1

96

1.1

1.0

Using molecular sieves Beds of molecular sieves are used to recover a solvent with 0.1%. This process requires around 2-2.5 kWh heat/kg of recovered solvent. Regeneration loop pressure drops are high, because of the additional molsieve beds and valves. Separate condensation of water Initially, when regenerative heating of the bed starts, a very little solvent is desorbed, but much of the water (about 87%) is desorbed. The circulating gas is first passed through a cooler and then a separate chiller for the condensation or freezing of the water. Subsequently, the gas is passed to the chiller where the water is condensed or frozen. From 1 to 1.8% water is recovered. This process is simpler than the molecular sieves process. Two separate chillers Two separate chillers are used in another adsorption process which offers further substantial improvements. A third activated carbon bed is added so that while two of the beds are being regenerated one of these is being cooled and transferring its heat to the other bed which is being heated. This not only gives excellent heat recovery but also provides a means of reducing the water content of the recovered solvent. The cooled bed absorbs the water vapor from the gas stream coming from the chiller. This dry gas stream is transported to the bed which is being heated and the desorbed solvent from the heated bed remains dry. The solvents recovered at this stage contain only 0.4 to 0.5% water. Solvent which is removed from the bed being cooled is readsorbed on the bed being heated. The process, which is patented, brings important benefits of regeneration gas flow rates. Such low flow rates mean that the process requires less heat, less refrigeration, and a

Klaus-Dirk Henning

1665

lower cost regeneration loop. The overall efficiency is high because the low residual solvent loading of the regenerated beds leads to increased solvent recovery. Two separate chillers plus molecular sieve bed A two chiller system with a molecular sieve bed has to be regenerated only about once per week since most of the water is removed in the chiller and readsorbed on the cooled carbon bed. This process offers a high solvent recovery rate giving a solvent with only 0.1% water at greatly reduced heat consumption, lower investment, and higher solvent recovery rate. Key plants with all the above-mentioned processes are offered by special engineering companies.31,32 The investment cost for a typical recovery plant with hot gas regeneration can be estimated by the following equation: COST (in 1000 Euro) = 200 + 0.4 × A0.7 +15 × S0.7 where: A S

feed air rate in Nm3/h solvent flow rate in kg/h

21.1.5.2.2 Solvent recovery with adsorption wheels Adsorption wheels33 are used for the continuous purification of large volumes of exhaust air containing relatively low solvent concentrations. The adsorption wheel consists of a number of identical chambers arranged axially around a vertical axis. All chambers contain adsorbent. The wheel rotates and each chamber passes in sequence over an exhaust air duct and solvent molecules are adsorbed. As wheel continues to rotate, the chamber in which adsorption had occurred now moves into a desorption position in which hot air is passed through the chamber and over the adsorbent. This removes the adsorbed solvent. The hot air flow in the desorption sector is at relatively low flow rate compared to that of the contaminated gas stream. The number of chambers in the desorption zone is much greater than the number in the absorption zone so the desorption stream is many times more concentrated in the solvent than was the exhaust stream. At this higher concentration, the desorption stream can now be economically treated in one of the ways: • The adsorption stream is purified by means of recuperative or regenerative oxidation. This approach is advantageous in painting applications with solvent mixtures that cannot be reused in production. • For solvent recovery by condensation, the desorption stream is cooled down in a cooling aggregate and the liquid solvent is recovered for reuse. Water contents below 1% can be achieved without further purification. Example: A manufacturer who specializes in flexible packaging, i.e., for confectionary, always adds the same solvent mixture (ethanol, ethyl acetate, ethoxypropanol) to his printing ink. Adsorptive solvent removal of the solvent mixture by use of an adsorption wheel (Figure 21.1.20) and solvent recovery via condensation has been technically and economically proven. 10,000 to 65,000 Nm3/h of exhaust air from printing machines and washing plants at a temperature maximum of 45°C and a maximum solvent loading of 4.6 g/m3 is

1666

21.1 Absorptive Solvent Recovery

to be cleaned. Depending on the air volume two adsorption wheels with a capacity of 26,000 Nm3/h and 39,000 Nm3/h and separate desorption circuits are used either alternatively or together. The total desorption air of max. 11,000 Nm3/h is being concentrated to max. 27 g/m3, which corresponds to 50% of the lower explosion limit. The condensation unit for the solvent recovery process is gradually adjusted to small, medium or large exhaust air volumes. The recovered solvents are stored and returned to the production process.

Figure 21.1.20. Adsorption wheel for solvent recovery in packaging printing [After reference 33].

21.1.5.3 Viscose industry In plants which produce viscose fiber (stable fiber), viscose filament yarn (rayon) and viscose film (cellophane), large volumes of exhaust air contaminated with carbon disulfide (CS2) and hydrogen sulfide (H2S) have to be cleaned. Typical waste gas volumes and concentration of CS2 and H2S are given in Table 21.1.11. Table 21.1.11. Waste gas in the viscose industry [After reference 35]. Product Rayon

Spec. waste gas volume, m3 400,000-700,000

Concentration (mg/m3) H2S

CS2

60-130

300-800

Klaus-Dirk Henning

1667

Table 21.1.11. Waste gas in the viscose industry [After reference 35]. Product Staple fibre Viscose fibre

Spec. waste gas volume, m3

Concentration (mg/m3) H2S

CS2

50,000-90,000

700-1800

2300-4000

100,000-150,000

280-400

1000-3000

Different cleaning/recovery processes are available for the removal of each of the two sulfur-containing compounds:2,17,34,35 Hydrogen sulfide Absorption of H2S in a NaOH-scrubber Catalytic oxidation of H2S to elemental sulfur on iodine impregnated activated carbon (Sulfosorbon-process) 2H 2 S + O 2 → 2S + 2H 2 O The sulfur is adsorbed at the internal surface of the carbon-catalyst. Carbon disulfide Adsorptive removal on activated carbon and recovery by steam desorption. For simultaneous H2S and CS2 removal, the Sulfosorbon-process uses adsorbers packed with two different activated carbon types • Iodine impregnated wide-pore activated carbon for H2S oxidation in the bottom part (gas inlet) of the adsorber • Medium-pore activated carbon for the adsorption of CS2 in the upper layer of the fixed bed. As soon as the CS2 concentration in the treated air approaches the emission limit, the exhaust air stream is directed to a regenerated adsorber. After an inert gas purge, the carbon disulfide is desorbed with steam at 110 to 130°C. The resulting CS2/steam mixture is routed through a condenser and cooler, before entering a gravity separator where phase separation occurs. At the usual CS2 and H2S concentrations in viscose production exhaust air, the CS2 adsorption/desorption cycles can be run for a prolonged periods before the first layer loaded with elemental sulfur has to be regenerated. The regeneration process of the sulfur loaded activated carbon involves the following steps: • washing out of sulfuric acid with water • extraction of elemental sulfur with carbon disulfide • desorption of carbon disulfide with steam • air drying and cooling of activated carbon The elemental sulfur present in the carbon disulfide in dissolved form can be separated by distillation and recovered as high-purity sulfur. Design example34 In the production of viscose filament yarn, a Supersorbon®-system combined with a NaOH-scrubber system (Figure 21.1.21) has been in use since 1997. A very high standard

1668

21.1 Absorptive Solvent Recovery

Figure 21.1.21. H2S and CS2 removal in the filament yarn production [After reference 34].

of safety engineering is implemented in the treatment system owing to the flammability of the CS2 and the toxic nature of the constituents to be removed from the exhaust air. System concept First treatment stage: Absorption of H2S in two NaOH jet scrubbers and one water-operated centrifugal scrubber. Second treatment stage: Fixed-bed adsorption for purification of exhaust air and CS2 recovery. Treated air stream Exhaust air from viscose filament yarn plant. Design data: Exhaust air flow rate 12,000 Nm3/h Solvent capacity 27 kg CS2/h and 7.5 kg H2S/h Exhaust air temperature 15-30°C Clean air solvent concentration 100 mg/Nm3 (24-hour mean) CS2: max. 5 mg/Nm3 (24-hour mean) H2S: max. H2S-removal The exhaust air, saturated with water vapor, is first treated in an absorption unit for removal of H2S by routing it through two successive jet scrubbers working with a dilute caustic soda solution. Downstream of these, a centrifugal scrubber is installed as an

Klaus-Dirk Henning

1669

entrainment separator. The sulfide-containing solution rejected from the scrubbers is used to precipitate zinc in the wastewater treatment system of the production plant. By-products adsorbed on the activated carbon, such as sulfuric acid and elemental sulfur, are removed periodically by water and alkaline extraction. CS2-removal and recovery The pre-cleaned airstream containing carbon disulfide vapors passes through two or three parallel adsorbers, in which the CS2 is absorbed on a bed of Supersorbon® activated carbon. As soon as the adsorbent is saturated it is regenerated by desorbing the solvent with a countercurrent flow of steam. The resulting mixture of water and CS2 vapors is condensed and separated by gravity settlers. The recovered CS2 is returned to the viscose production without further treatment. The condensed steam is stripped of residual CS2 in the centrifugal scrubber and used as dilution water for operation of the jet scrubbers. Operating experience The specified purity of the treated exhaust air is reliably achieved with the Supersorbon® process34 with respect to both CS2 and H2S. The CS2 recovery rate is about 95%. The purity of the recovered solvent meets viscose production specifications. A widely varying CS2 concentration in the exhaust air has not adversely affected operation of the adsorption system. 21.1.5.4 Refrigerator recycling Condensation process Condensation processes are especially suitable for the cleaning of low flow highly concentrated streams of exhaust gas.36 The entire waste gas stream is cooled below the dew point of the vapors contained therein so that these can condense on the surface of the heat exchanger (partial condensation). Theoretically, the achievable recovery rates depend only on the initial concentration, the purification temperature and the vapor pressure of the condensable at that temperature. In practice, however, flow velocities, temperature profiles, the geometry of the equipment, etc. play decisive roles, as effects such as mist formation (aerosols), uneven flow in the condensers and uncontrolled ice formation interfere with the process of condensation and prevent an equilibrium concentration from being reached at the low temperatures. The Rekusolv process37 uses liquid nitrogen to liquefy or freeze vapors contained in the exhaust gas stream. In order to reduce the residual concentrations in the exhaust to the legally required limits, it is often necessary to resort to temperatures below −100°C. The Rekusolv process is quite commonly used in the chemical and pharmaceutical industry and at recycling plants for solvent recovery. Example of solvent recovery in refrigerator recycling36,37 In refrigerator recycling plants R11 or pentane is released from the insulating foam of the refrigerator during shredding. Figure 21.1.22. shows the Rekusolv process as it is used by a refrigerator recycling company. At this plant, around 25 refrigerators per hour are recycled thereby generating 8 kg/h of polluting gases. The Rekusolv plant is capable of condensing almost all of this. The unit is designed to operate for 10 to 12 hours before it has to be defrosted. The plant is operated during the day and is automatically defrosted at

1670

21.1 Absorptive Solvent Recovery

Figure 21.1.22. Block diagram of the Rekusolv process [After references 36,37].

night. The concentration of pollutants in the exhaust gas is reduced from 20 to 40 g/m3 to 0.1 g/m3 − a recovery rate of more than 99.5%. 21.1.5.5 Petrochemical industry and tank farms Vapor recovery units are installed at petrochemical plants, tank farms and distribution terminals of refineries. Tank venting gases are normally small in volume, discharged intermittently at ambient temperature and pressure and loaded with high concentrations of organic vapors.2 Processes used for the cleanup of such waste air streams with organic vapor up to saturation point are often combined processes:2,38,39 • absorption and pressure-swing adsorption • membrane permeation and pressure-swing adsorption • condensation and adsorption Adsorption on activated carbon and vacuum regeneration Figure 21.1.23 represents the basic principle of the systems which are used to recover vapors displaced from tank farms and loading stations and blend them back into the liquids being loaded. The plant consists of two fixed-bed adsorbers packed with acti-

Klaus-Dirk Henning

1671

Figure 21.1.23. Vapor recovery by adsorption and vacuum regeneration [After reference 39].

vated carbon which operate alternatively. The vapor pressure gradients required as a driving force for desorption is generated by vacuum pumps. As the raw gas entering the adsorber is saturated with hydrocarbon vapors, the heat of adsorption causes local heating of the well-insulated adsorbent bed. The higher temperature of the spent activated carbon supports subsequent desorption as the pressure is being lowered by means of a vacuum pump. The desorbed vapor is transported from the adsorption bed via a vacuum pump. A liquid ring pump using water for the seal fluid is normally applied for safety purposes in operations where the air is allowed to enter the adsorbers. The desorption stream is processed through a cooler and absorber. The absorber uses gasoline (or the recovered solvent) as a sorbent and serves as a hydrocarbon reduction stage for the entire system. Loading facilities with the inlet vapor concentrations of over 10% by volume can demonstrate recovery efficiencies of over 99%. 21.1.5.6 Chemical industry In the chemical industry, a variety of solvents and solvent mixtures are in use. For many gas cleaning operations, special adsorption processes are required. Two-stage adsorption40 In many cases, a two-stage process offers significant advantages. This process makes use of two adsorbers that are operating alternately in series. Four-way reverse valves ensure safe changeover. During operation, the air passages of the valves are always open so that any blocking of the flow is impossible. Solvent laden air first passes into one of the

1672

21.1 Absorptive Solvent Recovery

adsorbers. Purified air is released into the atmosphere only during the time when the second adsorber is being steamed. During this phase of the process, the on-line adsorber is only slightly charged with a solvent which means that it will be completely adsorbed. In general, the time required for steaming an adsorber is at most 50% of the time provided for adsorption. After the adsorber has been steamed, the outlet valve is switched over. The steamed adsorber is then dried with solvent-free air taken from the laden adsorber. A feature of this process is that the steamed adsorber is always dried with solvent-free air. An advantage of this two-stage process is that the laden adsorber can be loaded beyond the breakthrough capacity since the second adsorber is unladen and can, therefore, accept any excess solvent. The condensation phase and reprocessing of the desorbate are carried out in the same manner as described for the single-stage plant. For example, a two-stage adsorption plant for the recovery of dichloromethane is able to clean a gas stream of 3000 m3/h to a residual solvent content of less than 20 mg/m3. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

J.W. Leatherdale, Air Pollution Control by Adsorption, Carbon Adsorption Handbook, Ann Arbor Science Publisher Inc., Michigan, 1978 P.N. edited by Cheremisinoff, F. Ellerbusch. VDI 3674, Waste gas cleaning by adsorption, Process and waste gas cleaning, Beuth Verlag GmbH, April, 2013. H. Menig, Luftreinhaltung durch Adsorption, Absorption und Oxidation, Deutscher Fachschriften-Verlag, Wiesbaden, 1977. H. von Kienle, E. Bäder, Aktivkohle und ihre industrielle Anwendung, Ferdinand Enke Verlag, Stuttgart, 1980. Test methods for activated carbon. European Council of Chemical Manufacturers` Federations/CEFIC Brussels (1986) 7. A. Yehaskel, Activated Carbon, Manufacture and Regeneration, Noyes DATA Corporation, Park Ridge, New Jersey, USA, 1978. T. Bandosz, Activated Carbon Surfaces in Environmental Remediation, Elsevier, 2006. M. Smisek, S. Cerny, Activated Carbon. Elsevier Publishing Company, Amsterdam-London-New York, 1970. H. Marsh, F. Rodriguez-Reinoso, Activated Carbon, Elsevier, 2006. R.Ch. Bansal, J.-B. Donnet and F. Stoeckli, Acitvated Carbon, Marcel Dekker Inc., New York, 1988. H. Jüntgen, Grundlagen der Adsorption, Staub-Reinhaltung Luft, 36, (1976) No. 7, pp. 281-324. H. Seewald, Technische verfügbare Adsorbentien. Vortragsveröffentlichung Haus der Technik, H. 404, pp. 24-34, Vulkan-Verlag, Essen, 1977. W. Kast, Adsorption aus der Gasphase. VCH-Verlag, Weinheim, 1988. H. Krill, Adsorptive Abgasreinigung - eine Bestandsaufnahme, VDI-Berichte 1034, pp. 339-371, VDI-Verlag, Düsseldorf, 1993. K.-D. Henning, Activated Carbon for Solvent Recovery, Company booklet CarboTech Aktivkohlen GmbH, D-45307 Essen, Germany. A. Mersmann, G.G. Börger, S. Scholl, Abtrennung und Rückgewinnung von gasförmigen Stoffen durch Adsorption, Chem.-Ing.-Techn., 63, No. 9, pp. 892-903. H. Jüntgen, Physikalisch-chemische und verfahrenstechnische Grundlagen von Adsorptionsverfahren, Vortragsveröffentlichung Haus der Technik, H. 404, pp. 5-23, Vulkan-Verlag, Essen, 1977. K.-D. Henning, Science and Technology of Carbon, Eurocarbon, Strasbourg/France, July 5-9, 1998. K.-D. Henning, S. Schäfer, Impregnated activated carbon for environmental protection, Gas Separation & Purification, 1993, Vol. 7, No. 4, p. 235. P.N. Cheremisinoff, Volatile Organic Compounds, Pollution Engineering, March 1985, p.30 C.S. Parmele, W.L. O`Connell, H.S. Basdekis, Vapor-phase cuts, Chemical Engineering, Dec. 31, 1979, p. 59.

Klaus-Dirk Henning 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39

1673

M.G. Howard, Organic solvents recovered efficiently by dry processes, Polymers Paint Colour Journal, May 18, 1983, p. 320. W.D. Lovett, F.T. Cunniff, Air Pollution Control by Activated Carbon, Chemical Engineering Progress, Vol. 70, No. 5, May 1974, p. 43. H.L. Barnebey, Activated Charcoal in the Petrochemical Industry, Chemical Engineering Progress, Vol. 67, No. 11, Nov. 1971, p. 45. K.-D. Henning, W. Bongartz, J. Degel, Adsorptive Removal and Recovery of Organic Solvents, VfT Symposium on Non-Waste Technology, Espo/Finnland, June 20-23, 1988. G.G. Börger, A. Jonas, Bessere Abflugtreinigung in Aktivkohle-Adsorben durch weniger Wasser, Chem.-Ing.-Techn., 58 (1986), p. 610-611, Synopse 15.12. K.-D. Henning, W. Bongartz, J. Degel, Adsorptive Recovery of problematic solvents, Paper presented at Nineteenth Biennial Conference on Carbon, June 25-30, 1989, Pennstate University/Pennsylvaia/USA. A.A. Naujokas, Loss Preventation, 12 (1979) p. 128. G. Konrad, G. Eigenberger, Rotoradsorber zur Abluftreinigung und Lösemittelrückgewinnung. Chem.-Ing.-Techn., 66 (1994) No. 3, pp. 321-331. H.W. Bräuer, Abluftreinigung beim Flugzeugbau, Lösemittelrückgewinnung in der Wirbelschicht, UTA Umwelt Technologie Aktuell (1993), No. 2, pp. 115-120. W.N. Tuttle, Recent developments in vacuum regenerated activated carbon based hydrocarbon vapour recovery systems, Port Technology International, 1, pp. 143-146, 1995. Exhaust air purification and solvent recovery by the Supersorbon(r) Process, Company booklet, Lurgi Aktivkohlen GmbH, D-60388 Frankfurt, Gwinnerstr. 27-33, Germany. G. Caroprese, Company booklet, Dammer s.r.l., I-20017 Passirana die Rhol (Mi), Italy. R. Peinze, Company booklet, Air Engineering, CH-5465 Mellikon, Switzerland. Environmental Technology, Company booklet, Eisenmann, D-71002 Böblingen, Germany. Exhaust air purification and CS2 recovery by the Supersorbon® process with front- and H2S absorption, Company booklet, Lurgi Aktivkohlen GmbH, D-60388 Frankfurt, Germany. U. Stöcker, Abgasreinigung in der Viskose-Herstellung und -Verarbeitung, Chem.-Ing.-Techn., 48, 1976, No. 10. Exhaust gas purification, Company booklet, Messer Griesheim GmbH, D-47805 Krefeld, Germany. F. Herzog, J. Busse, S. Terkatz, Refrigerator recycling - how to deal with CFCs and pentane, Focus on gas 13, Company booklet, Messer Griesheim GmbH, D-47805 Krefeld, Germany. C.J. Cantrell, Vapour recovery for refineries and petrochemical plants, CEP, October 1982, pp. 56-60. Kaldair Vapor Recovery Process, Company booklet, Kaldair Ltd., Langley Berks SL3 6EY, England.

1674

21.2 Recovery Versus Incineration

21.2 RECOVERY VERSUS INCINERATION *Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa *INGAR (CONICET), Avellaneda 3657, S3002GJC Santa Fe, Argentina ** Department of Environment, Land and Infrastructure Engineering (DIATI), Politecnico di Torino, Corso Duca degli Abruzzi 24, 10129 Torino, Italy

21.2.1 INTRODUCTION Organic solvents are widely used in the pharmaceutical industry when performing, among several tasks, the reaction, separation, and purification steps involved in the manufacture of a drug product. As demonstrated by Luis et al.,1 life cycle assessment (LCA) can be used as a tool to determine the best waste solvent technology from an environmental point of view and even as an a priori decision making tool.2 Incineration is the most obvious option when considering the use of waste solvents as fuels for steam and electricity production. Waste solvent incineration and incineration in a cement kiln are alternatives that deserve attention in such a case.3,4 Solvent recovery, on the other hand, minimizes the usage of raw materials. Batch distillation is the key recovery operation because high recovery and purity can typically be achieved. In addition, solvents from different waste streams can be recovered using the same column turning the recovery system both financially advantageous as well as environmentally friendly.5 However, it is noteworthy that hybrid processes comprising distillation and membrane separation technologies like pervaporation are emerging as suitable choices, mostly due to the independence of the separation performance of membrane technologies from the phase equilibrium. Hence, the hybrid process is able to break azeotropes of azeotrope-forming waste streams while reducing the energy demand of the distillation task. Moreover, separations by distillation that are controlled by tangent pinch points are also candidates to be replaced by a hybrid process since recovering the solvent in high purity by distillation alone would require not only high instantaneous reflux ratios but also demand a significantly high number of theoretical trays. In such a case, distillate purity control is also an issue due to the occurrence of a zone of constant composition near the rectifier top.6 Urtiaga et al.7 demonstrated the technical feasibility of isopropyl alcohol recovery with the hybrid distillation/pervaporation. Koczca et al.8 performed economic comparisons between the state-of-the-art and hybrid alternatives in the recovery of tetrahydrofuran from binary and ternary mixtures demonstrating the economic benefits of hybrid processes including pervaporation. Slater et al.9 proposed the integration of pervaporation with a constant volume distillation for the recovery and reuse of tetrahydrofuran. The aforementioned authors assessed the effectiveness of the hybrid technology resorting to an analysis taking into account both economic and environmental aspects. Applied to the mixture THF-methanol, Luis et al.10 performed a thorough analysis of the alternatives pressure swing distillation, distillation/pervaporation, and incineration in terms of their environmental impact (LCA) and energy consumption (simulation), concluding that the hybrid process can be considered a real alternative to the other technolo-

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1675

gies due to the saving in materials and energy in an overall scenario. Investment costs were not included, however, in their analysis. In this chapter, the economic and environmental performance of the alternatives distillation/incineration and distillation/pervaporation for the treatment of the azeotropic mixture isopropyl alcohol-water, whose separation by distillation is characterized by an energy demand controlled by the tangent pinch points, are summarized.11 The incineration alternative, either in a conventional waste solvent incineration plant or in a cement kiln, resorts to a pre-concentration step by distillation. The recovery alternative, on the other hand, considers a hybrid process consisting of a distillation column followed by a pervaporation unit. Quasi-optimal designs taking into account both operating and investment costs for each technological alternative were obtained resorting to the conceptual models of the units involved.12,13 Optimal values for the operation variables obtained at the conceptual design level were used as input data of the life cycle analysis to consider all impacts on humans and the environment during the entire life cycle of the solvent.5,14,15 21.2.2 DESCRIPTION OF THE TREATMENT ALTERNATIVES FOR A CASE STUDY Two process alternatives for the treatment of 5 t/day of a binary mixture isopropyl alcoholwater with 27 wt% of isopropyl alcohol (IPA) will be analyzed. For alternative A1, whose operating sequence is depicted in Figure 21.2.1(a),16 the fresh feed is concentrated via batch distillation to obtain a distillate with a composition near that of the azeotrope IPA-water. After that, while the distillate is sent to off-site incineration in a cement plant, the column residue is diverted to a biological disposal facility

Figure 21.2.1. Operation sequence corresponding to: (a) Alternative A1, the first cut is sent to off-site incineration, while the second cut is recycled to the next batch, (b) Alternative A2, the first cut is treated in the pervaporation unit while both the second cut and the permeate of the membrane unit are recycled to the batch rectifier in the next batch. [Adapted, by permission, from A. Amelio, D.A. Figueroa Paredes, J. Degrève, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual Model-Based Design and Environmental Evaluation of Waste Solvent Technologies: Application to the separation of the mixture acetone-water, Sep. Sci. Technol., 53 (11), 1791, (2018).]

1676

21.2 Recovery Versus Incineration

on-site. For this alternative, it is considered that the optimal process consists of concentrating the fresh feed up to a concentration near the azeotropic composition (88.5 wt% IPA) to obtain a mixture with a lower calorific value above 5500 kcal/kg. In this way, support fuel is avoided in the incineration task, and savings are expected in haulage costs due to the decreased volume to be transported from the pharmaceutical plant to the cement kiln.6 It is assumed that the distillation column can operate 20 hours a day, consuming the remaining 4 hours for filling, starting-up and emptying the column. The primary optimizaD tion variable is the “per-pass” recovery of alcohol, σ IPA . in the first cut. Table 21.2.1 shows the lower calorific value corresponding to a mixture with 83 wt% of isopropyl alcohol, which is adopted as the target distillate composition during the first cut. The first cut, which is commonly operated at a variable reflux policy, is followed by a second cut operated at a constant reflux ratio policy (i.e., 0.5) to achieve an IPA concentration below 1 wt% in the boiler at the end of the batch. The second cut is recycled to the next batch to avoid IPA losses, as shown in Figure 21.2.1(a). Main process specifications for alternative A1 are summarized in Table 21.2.2. Table 21.2.1. Lower calorific value of the concentrated feed (83 wt% IPA). [Data from reference 11]. Heat of combustion IPA [Kcal/kg]

7201

Latent heat W [kcal/kg]

539

Lower calorific value 83% IPA

5976.83

Lower calorific value 17% W

-91.63

Lower calorific value 83:17 Mixture

5882.20

Table 21.2.2. Process specifications for the treatment alternatives. Treatment alternatives

Distillation/incineration (A1)

Distillation/pervaporation (A2)

Fresh feed amount and IPA weight percent

5 t/day, 27 wt%

5 t/day, 27 wt%

Average distillate composition (first cut)

mixture with a lower calorific value above 5500 Kcal/Kg

value subject to optimization taking into account operating and investment costs

IPA weight percent in the boiler at the end of the distillation task

< 1 wt%

< 1 wt%

IPA weight percent in the product tank at the operation end

-

> 99.8 wt%

For alternative A2, on the other hand, the alcohol is recovered in a two-step process comprised by a distillation column and a hydrophilic pervaporation unit, both units operated in batchwise mode. As in alternative A1, the residue of the column is sent to biological treatment and discharge. Figure 21.2.1(b) shows the operating sequence corresponding to alternative A2. The sequence for the distillation column is similar to that of alternative

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1677

Figure 21.2.2. “Scheduling” corresponding to alternative A2.

A1. The first cut is sent to the membrane unit, which is also operated in batchwise mode. The isopropyl alcohol-rich distillate stored in the product tank is enriched in its IPA concentration until achieving an alcohol composition above 99.8 wt%. In this case, the dehydrated IPA can be recycled to the main pharmaceutical process thus minimizing the use of a fresh solvent. Note that both the second cut and the permeate from the membrane unit must be recycled to the batch rectifier in the next batch to avoid isopropyl alcohol losses. Obtaining a quasi-optimal design for this alternative is more complex than for the alternative distillation/incineration since values for an increased number of optimization variables must be determined. Main optimization variables are: i) “per-pass” recovery of D D alcohol σ IPA in the first cut, ii) distillate mole fraction, x IPA , in the first cut, iii) distillation processing time, tD, iv) pervaporation processing time, tM, and v) permeate pressure, Pp. Note that by parametrically varying the distillate mole fraction of the first cut, which is operated at a variable reflux policy, allows capturing the economic trade-offs between distillation and membrane tasks. To narrow the search space, it is assumed that the reflux ratio of the second cut is 0.5. The permeate pressure is 1.5 kPa.7 Operating times for distillation and dehydration tasks are set to 16 and 20 hours, respectively, according to the tasks schedule shown in Figure 21.2.2.16 The cycle time is 24 hours with a time-horizon of 360 days/year. Pseudo-steady state of the process is typically achieved after a relatively low number of cycles. Main process specifications for alternative A2 are summarized in Table 21.2.2. 21.2.3 CONCEPTUAL MODELING OF THE TREATMENT ALTERNATIVES In the context of optimization-based design, Marquardt et al.17 proposed a three-step approach for the design of hybrid processes: (i) generation of flow-sheet alternatives, (ii) evaluation of the alternatives with the aid of shortcut methods; (iii) rigorous optimization of the most promising alternatives to obtain the best flow-sheet. In this line, Caballero et al.18 successfully applied the last two steps of the mentioned approach in the optimization of a hybrid separation system composed of a distillation column and a vapor permeation unit. Considering detailed rate-based models for distillation and pervaporation, Koch et al.19 optimized a superstructure of the hybrid process using evolutionary algorithms for the simultaneous determination of the optimal process configuration, equipment design,

1678

21.2 Recovery Versus Incineration

and operating conditions. To bridge the gap between shortcut calculations and detailed engineering design, Skiborowski et al.20 proposed an efficient and robust approach for the optimal design of membrane-assisted distillation processes. Meyer et al.11 and Amelio et al.16 applied the second step of the approach proposed by Marquardt et al.17 to the evaluation of alternatives for the treatment of solvents. Given that flow-sheets for alternatives A1 and A2 are defined according to Figure 21.2.1, the methodology considered in this chapter intends to show how the economic and environmental performance of both waste solvent technologies can be evaluated from the conceptual design of each process. This task was performed resorting to conceptual models or shortcuts for each operation unit involved, models that are characterized by their abilities to capture the essence of the process at low computational cost. Considering an additional cost model including both investment and operating costs, this approach is suitable for the optimization-based design of waste solvent technologies. Quasi-optimal values for the utilities, that is, cooling water, electricity, and steam for distillation and pervaporation are, on the other hand, the input data necessary to perform the environmental assessment.5,11,16 Main issues in the conceptual modeling of each unit are highlighted in the following sections. 21.2.3.1 Conceptual model for the distillation task For the case study under consideration, the quasi-optimal operation of the batch rectifier was estimated based on a conceptual model of the operation. The main assumptions of the conceptual model are: (i) the rectifier has an infinite number of stages, and (ii) the instant variations of the molar hold-up in the trays are negligible. Under these assumptions, the following relationships apply for every component in the mixture:21 D

D

xi dσ i ---------- = -----B dη x i0

[21.2.1] D

B B ( 1 – σi ) x i = x i0 ------------------(1 – η)

where:

[21.2.2]

D

σ i fractional recovery of component i in the distillate rectification advance η D x i mole fraction of component i in the distillate B

x i0 initial mole fraction of component i in the still

Equation [21.2.1] above has the advantage of considering the component recoveries and the rectification advance instead of the component mole fractions. This choice is appropriate at the design level considering that the design specifications in batch distillation are normally given in terms of component recoveries and that the vapor flow rate does not need to be specified at this level. Note that for given instantaneous values of the rectification advance and the recovery of each component in the distillate, it is possible to calculate the instantaneous still composition from Equation [21.2.2]. This equation gives the

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

Figure 21.2.3. Schematics of an instantaneous separation with two controlling pinch points: a node pinch at the bottom and a saddle tangent pinch in the “middle” of the rectifier. [Data from reference 22].

1679

instant still composition needed for calculating the instant distillate composition (if the column is operated at constant reflux) or for calculating the instantaneous reflux ratio (if the column is operated at constant distillate composition). Focusing our attention on the operation at constant reflux ratio, the conceptual model must predict the instantaneous distillate composition with the aid of pinch theory.22 The composition of the controlling pinch point must be determined to perform the mass balance around an envelope relating either a node pinch (Figure 21.2.3, Envelope I) or a tangent pinch (Figure 21.2.3, Envelope II) with the unknown distillate composition. Although Figure 21.2.3 shows a special case where two pinch points control the separation, the analysis of the Figure enhances comprehension of the basics of the conceptual model. As usual, equimolal overflow is assumed, and hence, the operating lines take the form of straight

lines in the y-x diagram. At pinch situation the mass balance around either Envelope I or II becomes: xD r * - = 0 – y x + ----------- x p + ---------p r+1 r+1

[21.2.3]

where: xp y*xp xD r

pinch point lying on the y-x equilibrium curve vapor in phase equilibrium with xp distillate composition reflux ratio

In the case of a node pinch controlling the separation, xp = xD and hence, Equation [21.2.3] may be solved for the mole fraction of the distillate composition xD from known values of the pinch xp and the reflux ratio r. Otherwise, there is an additional constraint on the slope of the equilibrium curve at the tangent pinch that must be taken into account:23

1680

21.2 Recovery Versus Incineration

dy -----dx

r = ----------r+1

x = xp

[21.2.4]

with the right-hand side of Equation [21.2.4] representing the slope of the operating line of the rectifier. In this case, Equations [21.2.3] and [21.2.4] have to be solved simultaneously for the mole fraction of the distillate composition xD and the tangent pinch composition xp. Different algorithms needed for estimating the instantaneous rectifier performance at different reflux policies can be found in Torres and Espinosa.22 Note that for the case study, the equilibrium curve shows both an azeotrope and an inflexion point IP, which is typical in systems where the energy demand is controlled by tangent Figure 21.2.4. Diagram y vs. x for IPA-water mixture at pinch points for certain separations, as is 101.3 kPa. IP represents the inflection point of the equi- shown in Figure 21.2.4. In the case of startlibrium curve. [Adapted, by permission, from R. Meyer, ing compositions at the left of the inflection D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. point, the choice of distillate compositions Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, crit Conceptual model-based optimization and environmen- lower than x D will result in instantaneous tal evaluation of waste solvent technologies: Distillaseparations for which the energy demand is tion/incineration versus distillation/pervaporation, Sep. controlled by the composition in the boiler; Purif. Technol., 158, 238 (2016)]. that is, operations with similar behavior to crit ideal mixtures. For distillate compositions higher than x D , the instantaneous minimum reflux can be controlled either by a "pinch" at the bottom of the column, a tangent pinch located somewhere in the “middle”, or both.22 It is noteworthy that the conceptual model of the batch rectifier decouples the variation of compositions from the variation of flows and batch size. As stated by Bernot et al.,24 this kind of approach can be used to estimate batch sizes, operating times, equipment sizes, utility loads, and costs for the batch distillation. For given values of the batch size M0 [kmol] and operating time tD, the only variable that must be determined is the vapor flow rate V [kmol/h], which can be iteratively obtained from the results obtained at the conceptual model level and the mass balance around the condenser from: tD

η

1 1 V = ----  D ( t ) ( R ( t ) + 1 ) dt = ----  M 0 ( R ( η ) + 1 ) dη tD tD 0

where: V tD M0

vapor flow rate operating time batch size

0

[21.2.5]

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1681

For integrating Equation [21.2.5] the instantaneous reflux ratio was calculated from: R = f×r

[21.2.6]

where: f

factor

In all studied cases, the value of f was about 1.2. The number of trays was estimated 0 with the aid of DISTIL25 from ( x B , R(0)) to achieve the selected distillate composition xD for the first cut at the column top. The simulation of the instantaneous separation in ASPEN Hysys26 for the initial conditions was also used to determine both the column diameter and the type of internals. The tool Try Sizing (Aspen Hysys) recommended the adoption of a packing column with a diameter of 0.3048 m considering the relatively low vapor flow rates involved in all studies. Equation [21.2.5] was iteratively solved for different values of V until achieving the desired value of the operation time tD. Figure 21.2.5 depicts the time evolution of the primary variables calculated with the conceptual model for an IPA mole fraction of 0.65 in the first cut. The operation cycle of the batch rectifier is 16 h with a vapor flow rate of 5.359 kmol/h. The rectifier has 13 theoretical stages and a column diameter of 0.3048 m. For a case study analyzed in Amelio et al.,16 Figure 21.2.6 shows the time evolution of the reflux ratio obtained from rigorous simulation of the distillation task with ASPEN Batch Distillation.26 Whether the conceptual modeling approach, the rigorous simulation of the task or a combination of both approaches is used, two main issues must be taken into account: i) the selected model must be run until achieving pseudo-steady state and, ii) liquid-vapor equilibrium must be properly represented. The main advantage of the conceptual modeling approach is that the vapor flow rate and column diameter needed to perform the desired separation in the operating time tD are outputs of the model. Modeling the distillation task resorting to rigorous simulation, on the other hand, allows including the trays and condenser hold-up issue.27

1682

21.2 Recovery Versus Incineration

Figure 21.2.5. Time variation of (a) component recovery, (b) distillate composition, (c) still composition, (d) still Cut1 Cut1 and distillate temperatures, and (e) reflux ratio V = 5.359 kmol/h. x D = 0.65. x IPA = 0.92. [Data from reference 11].

21.2.3.2 Conceptual model for the pervaporation task The process of pervaporative separation consists of two steps: selective dissolution and diffusion of the components of a liquid charge through a dense membrane and its consequent partial evaporation. The driving force for permeation is the difference in chemical potential of components at the two sides of the membrane − a difference which is generally accomplished by lowering the pressure at the permeate side. A common dense membrane consists of a dense layer oriented towards the feed/retentate side in the

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1683

Figure 21.2.6. Time evolution of the reflux ratio obtained from rigorous simulations in Aspen Batch Distillation.26 [Data from reference 16].

Figure 21.2.7. Variation across the membrane of (a) retentate temperature, (b) permeate flux and, (c) water mole fraction in the retentate side for a quasi-optimal design. Mixture isopropyl alcohol-water. [Adapted, by permission, from M.A. Sosa, and J. Espinosa, Feasibility Analysis of Isopropanol Recovery by Hybrid Distillation/ Pervaporation Process with the Aid of Conceptual Models, Sep. Purif. Technol., 78, 237 (2011)].

pervaporation cell on top of a porous supporting layer oriented towards the permeate side in which vaporization occurs.28 In the typical arrangement of a staged pervaporation process, a heat exchanger is placed after either a constant temperature drop of the liquid mixture or a constant membrane area. The decrease in the temperature, which results in a decrease of the driving

1684

21.2 Recovery Versus Incineration

force for the permeation process, is due to the change of state of the permeating components, which take their vaporization heat from the retentate liquid. An additional drop in the driving force for the separation is caused by a concentration decrease along the module of the preferentially permeating component in the liquid mixture.29 The concepts above can be understood with the aid of Figure 21.2.7(a)-(c) that shows, for the mixture IPAwater, the variation of the retentate temperature, permeate flux, and water mole fraction in the retentate across a membrane unit composed by 5 modules of 15 m2 each, respectively.30 Results shown in Figure 21.2.7(a)-(c) were obtained from integration of a model of the membrane unit which incorporates the semi-empirical mass transfer model based on Fick’s law with concentration dependent diffusivity parameters developed by Urtiaga et al.7 for the commercial membrane CMC-CF-23 (cm-celfa Membrantrenntechnik AG). Table 21.2.3 shows the steps that must be followed to solve both mass and energy balances around the unit. Parameters needed to run the model can be found in Urtiaga et al.7 Table 21.2.3. Algorithm to solve mass and energy balances in the membrane unit. Mixture IPA-water.7,30 ____________________________________________________________________________________ f f f Step 1. Calculate components activities a i, 353.14 = x i γ i, 353.14 [21.2.7] ____________________________________________________________________________________ f Step 2. Calculate the water flux [kg/(m2h)] J w, 353.14 = A 1 [ exp ( ( A 2 a w ) – 1 ) ] [21.2.8] ____________________________________________________________________________________ 2 Step 3. Calculate the ratio of partial fluxes q and then the IPA flux [kg/(m h)]

J w, 353.14 A3 f q = -------------------------- = ------------ [ exp ( A 4 a w – 1 ) ] [21.2.9] f J IPA353.14 a ____________________________________________________________________________________ IPA

Step 4. Use the Arrhenius-type equation to account for the dependence of water and isopropyl alcohol fluxes [kg/(m2h)] with temperature E a, i J i, T = J i, T exp  – ----------  RT  0

[21.2.10]

where:

E a, i ln J i, T = ln J i, 353.14 + -------------------353.14R 0

[21.2.11] ____________________________________________________________________________________ Step 5. Calculate the molar fluxes [kmol/(m2h)] J i, T mol J i, T = ------------MW i

[21.2.12]

and then solve the mass balance corresponding to a differential area element from: mol

[21.2.13] d [ Lx i ] = – J i, T dA ____________________________________________________________________________________ Step 6. Solve the energy balance corresponding to a differential area element from: mol [21.2.14] Lc p dT = – ΔH vap J T dA ___________________________________________________________________________________ Step 7. Update values for the retentate flow rate, retentate mole fraction for each component and temperature and go to Step 1. Integrate the model until the selected module area is achieved. ____________________________________________________________________________________

Focusing our attention on the case study under consideration, the membrane system is formed by a feed tank, the membrane unit itself and a recirculation pump. For the sake of simplicity, the vacuum-refrigeration system is not shown in Figure 21.2.1(b). However, it must be optimized for each purification task; i.e., for different amounts and compositions of the first cut of the distillation task.

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1685

While the mass balance in the feed tank is modeled through a set of differential equations, the performance of the pervaporation unit of a given membrane area is calculated by integrating the spatially one-dimensional model shown in Table 21.2.3. The volumetric flow rate of the recirculation pump is fixed according to Reynolds numbers achievable in the module geometry chosen.11,16 Figure 21.2.8. Evolution of the temperature along the retentate side The final product is obtained in the of the membrane unit. Maximum operation temperature was set at feed tank at the operation end. As 90oC. [Adapted, by permission, from R. Meyer, D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der long as the feed to the membrane Bruggen, and J. Espinosa, Conceptual model-based optimization unit increases its alcohol content and environmental evaluation of waste solvent technologies: Distillation/incineration versus distillation/pervaporation, Sep. Purif. with time, the driving force for Technol., 158, 238 (2016)]. separation decreases as shown in Figure 21.2.8.11 Therefore, pervaporation modules should be arranged in parallel to maximize the driving force during the batch cycle. The model was implemented in the equation-oriented simulation software gPROMS (Process Systems Enterprise).31 21.2.3.3 Conceptual model for the vacuum production task Relevant aspects of the conceptual modeling of two different vacuum systems; namely, vacuum pump/refrigeration and steam-jet ejector systems, are briefly summarized in this section. 21.2.3.3.1 Vacuum pump/refrigeration system The vacuum pump/refrigeration system is formed by a condenser and a vacuum pump. The aim of this system is to recover the significant part of permeate as a liquid while decreasing the charge of the gas evacuated by the vacuum pump. The permeate stream entering to the condenser is first cooled from the operating membrane temperature to the dew point and then condensed until reaching the bubble point of the mixture, points which are a function of the pressure and the instantaneous composition of the permeate mixture. Given the relatively low value of the permeate pressure, the low temperature required for the permeate condensation is usually reached by using a refrigerant as a cooling service fluid. In this case, the condenser is coupled with a vapor compression refrigeration cycle using propane as refrigerant. Note that, inevitably, there is an inward leakage of air into the equipment under vacuum; namely, the membrane unit and the condenser. Given that the air remains always in the vapor phase, its mole fraction increases along the condenser. Particularly, at the condenser end, the vapor-liquid equilibrium is strongly affected by the presence of air so that the bubble point of the mixture is lower than that of the binary mixture. Another technical relevant aspect is the freezing point of the binary mixture, which depends on its composi-

1686

21.2 Recovery Versus Incineration

tion and acts as the lower bound of the operating temperature in the condenser to avoid solids accumulation in the heat-transfer area.32 For proper design of both the condenser and the vacuum pump, an appropriate description of the phase change of the mixture is performed by considering a multi-node model for the condenser.33 Given that inward leakage of air depends on the volume of the equipment under vacuum an iterative procedure is required to solve the material and energy balance together with the design equations for the condenser area and volume (including the membrane module).32 Optimal design of this sub-system is obtained by minimizing the overall annual cost of the process. In the objective function, a cost item accounting for the loss of non-condensed solvent is also considered. The optimization variables are the condensation temperature at the end of the condenser, and the refrigerant pressures in the vapor compression refrigeration cycle. 21.2.3.3.2 Steam-jet ejector system If the permeate stream has a low solvent content as is the case of highly selective pervaporation membranes it is not necessary to recover the solvent in the next batch of the distillation task.16 However, the vacuum level must still be maintained to allow the separation of the first cut of the distillation task into a solvent-rich product and a water-rich permeate. For this purpose, a conventional steam-jet ejector system can be adopted as an alternative with a lower investment cost than that of the vacuum pump/refrigeration system. Note, however, that the steam diluted permeate stream requires a biological treatment before discharge. To this end, it must be mixed with the still content at the end of the distillation task in the next batch. To model this system with the aid of shortcut models, the feed must be converted into an air equivalent in standard conditions.34 The data presented by Coker34 for steam and cooling water consumption must be correlated for a system formed by n ejectors with n-1 inter-condensers.16 21.2.4 ECONOMIC ANALYSIS The economic analysis can be done by applying the methodology explained in Seider et al.35 For the case study under consideration, it was performed by calculating the annualized investment cost and the annual operating cost of alternatives A1 and A2 for different values of the corresponding optimization variables (e.g., the “per-pass” recovery of IPA in the first cut of the distillation task for alternative A1 or the mole fraction of isopropyl alcohol in the first cut of the distillation task for alternative A2) until achieving the economic optimum for each treatment technology, expressed as an overall cost in U$S/t IPA. Table 21.2.4 shows the design and operation variables obtained at the conceptual design level for the alternative distillation/incineration, which in turn are the input data of the economic model described in detail in Meyer et al.11 Among the operating costs, the costs involved in transporting the concentrated distillate (Cut1 in Figure 21.2.1(a)) to the cement plant were estimated from parameters given in Table 21.2.5.11 Costs for both the incineration of the distillate and the biological disposal of the column residue (Residue2 in Figure 21.2.1(a)) together with costs related to the purchase of isopropyl alcohol and the consumption of utilities are shown in Table 21.2.6. Input data needed to estimate the eco-

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1687

nomic performance of the recovery alternative are summarized in Table 21.2.7, where design and operation variables corresponding to distillation and pervaporation tasks are taken into account. Note that the total capital investment for each alternative is annualized taking ten years as a lifetime with 10% interest rate. The annual utility consumption is calculated by considering the operating cycle time of each process unit in a time-horizon of 360 days/ year. Table 21.2.4. Design and operation variables for Alternative A1: Distillation/incineration. Condenser area [m2]

Packed diameter [m]

Evaporator area [m2]

Cooling water [m3/year]

Horizontal vessel volume [m3]

Steam [kg/year]

Packed height [m] Table 21.2.5. Parameters used to calculate the transport cost. [Adapted, by permission, from R. Meyer, D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual model-based optimization and environmental evaluation of waste solvent technologies: Distillation/incineration versus distillation/pervaporation, Sep. Purif. Technol., 158, 238 (2016)]. Truck capacity [m3/trip] Distance per trip [km/trip] Fuel economy [km/L] Fuel price [U$S/L] Average Truck Speed [km/h]

25 2 x220 5

Load/Unload time of products [h]

1

Driver wage [U$S/h]

10

Maintenance expenses [U$S/km]

0.0976

0.85

General expenses [U$S/d]

8.22

60

Truck: Purchase cost [U$S]

2E+05

Table 21.2.6. Costs for raw material, utilities, incineration and biological disposal. IPA [U$S/t]

1200

Electricity [U$S/MWh]

Cooling water [U$S/m3]

0.091

Incineration in cement kiln [U$S/t]

122.6 150

Steam (1 barg) [U$S/kg]

0.025

Biological disposal [U$S/t O2]

650

Table 21.2.7. Design and operation variables for Alternative A2: Distillation/pervaporation. Column 2

Membrane Unit 3

Condenser area [m ]

Vessel [m ]

Evaporator area [m2]

Vacuum pump duty [kW]

Horizontal vessel volume [m3]

Compressor duty [kW]

Packed height [m]

Recirculation pump flow rate [m3/h]

Packed diameter [m]

Permeate condenser area [m2]

3

Cooling water [m /year]

Retentate heater area [m2]

Steam [kg/year]

Propane condenser area [m2]

1688

21.2 Recovery Versus Incineration

Table 21.2.7. Design and operation variables for Alternative A2: Distillation/pervaporation. Column

Membrane Unit Membrane area [m2] Vacuum vessel volume [m3] Cooling water [m3/year] Steam [kg/year] Electricity [kWh/year]

Applied to the case of acetone-water described in detail in Amelio et al.,16 Tables 21.2.8 and 21.2.9 show the costs obtained from the optimization of alternatives A1 and A2 using the conceptual modeling approach together with a cost model. While for alternative A1 off-site incineration either in a cement kiln or a conventional waste solvent incinerator is considered, two different vacuum systems are taken into account for alternative A2. Note that in Table 21.2.9, the payback period and the internal rate of return IRR after ten years was calculated using a 35% income tax rate and a 10% interest rate to determine operating cost savings from solvent recovery.5 Table 21.2.8. Annualized investment and operating costs [103 U$S/year] for two variants of distillation/incineration. [Data from reference 16]. Variant Incineration in cement kiln

Conventional waste solvent incineration

Column investment cost

52

52

Column utilities cost

6.1

6.1

Column residue disposal cost

4.4

4.4

Transportation cost

10.7

1.7

Incineration cost

71.8

120

General cost

62

62

Overall cost

208

256

Acetone to incineration [t/year]

454

454

Overall cost [U$S/t Acetone]

457

563

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1689

Table 21.2.9. Annualized investment and operating costs [103 U$S/year] for two variants of distillation/pervaporation. [Data from reference 16]. Vacuum system variant Vacuum pump/refrigeration system

Steam-jet ejector system

Pervaporation unit investment cost

96.1

85.1

Column investment cost

52.7

52.2

Column utilities cost

6.4

6.1

Column residue disposal cost

5.8

7.1

Pervaporation unit utilities cost

1.9

5.4

Membrane replacement cost

3.9

3.9

General cost

116

110

Overall cost

283

270

Acetone recovered [t/year]

453

452

Overall recovery cost [U$S/t Acetone]

625

597

Payback period* [year]

3

3

Internal return rate* [%]

26

24

*annual earnings consider savings due to the avoided purchase of fresh solvent and the avoided disposal cost of the spent solvent.

Figure 21.2.9. System boundaries for the alternative distillation/incineration considering two incineration variants: (a) a Waste Solvent Incineration plant and (b) a cement kiln. [Adapted, by permission, from A. Amelio, D.A. Figueroa Paredes, J. Degrève, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual Model-Based Design and Environmental Evaluation of Waste Solvent Technologies: Application to the separation of the mixture acetone-water, Sep. Sci. Technol., 53 (11), 1791, (2018).]

1690

21.2 Recovery Versus Incineration

Figure 21.2.10. System boundaries for the alternative distillation/pervaporation considering two vacuum production variants: (a) a vacuum pump/refrigeration system and (b) a steam-jet ejector system. [Adapted, by permission, from A. Amelio, D.A. Figueroa Paredes, J. Degrève, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual Model-Based Design and Environmental Evaluation of Waste Solvent Technologies: Application to the separation of the mixture acetone-water, Sep. Sci. Technol., 53 (11), 1791, (2018)].

21.2.5 ENVIRONMENTAL ANALYSIS Life cycle assessment will be used to evaluate the environmental impact of the alternatives studied for one year of operation. Figure 21.2.9(a)-(b) shows the boundaries between the technological system and nature for alternative A1.16 Impacts related to the following tasks are considered for the distillation/incineration alternative: i) solvent production, ii) pre-concentration step, iii) transportation, iv) biological disposal and v) incineration either in a waste solvent incineration plant or in a cement kiln. All these steps bring a burden to the environment. It is noteworthy that for the waste solvent incineration variant benefits due to steam and electricity production must be taken into account. For the case of incineration in a cement kiln, on the other hand, benefits due to substituted fuels and their corresponding avoided emissions must be considered. Figure 21.2.10(a)-(b) shows the system boundaries for alternative A2.16 Impacts related to the following tasks are considered for the distillation/pervaporation alternative: i) solvent production, ii) pre-concentration step, iii) pervaporation, iv) biological disposal and v) vacuum production either with a vacuum pump/refrigeration system or a steam-jet ejector system. In this case, credits to the environment are given with the avoided production and incineration of the recovered solvent. The environmental burden related to the construction and installation of equipment is not within the LCA boundaries for none of the alternatives analyzed in this chapter, and, therefore, not included in the overall assessment.5

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1691

The environmental impact assessment was undertaken using the ReCiPe 200836 method from SimaPro 7.337 software package. ReCiPe 200836 includes eighteen midpoint categories and three endpoint categories as shown in Tables 21.2.10 and 11, respectively. While midpoints categories are accounted with their characterization factors (e.g., kt CO2 equivalent for the midpoint category climate change in Table 21.2.10), endpoints categories are presented regarding a single score.36,38 Table 21.2.10. Category indicators at the midpoint level.36 (1) climate change

(10) freshwater ecotoxicity

(2) ozone depletion

(11) marine ecotoxicity

(3) terrestrial acidification

(12) ionizing radiation

(4) freshwater eutrophication

(13) agricultural land occupation

(5) marine eutrophication

(14) urban land occupation

(6) human toxicity

(15) natural land transformation

(7) photochemical oxidant formation

(16) water depletion

(8) particulate matter formation

(17) mineral resource depletion

(9) terrestrial ecotoxicity

(18) fossil fuel depletion

Table 21.2.11. Category indicators at the endpoint level.36 (a) damage to human health (b) damage to ecosystem diversity (c) damage to resource availability

It must be emphasized that in order to carry out the environmental impact assessment, life cycle inventories corresponding to the tasks involved in alternatives A1 and A2 must be calculated from the results obtained at the conceptual modeling level on an annual basis and LCA software tools like Ecosolvent15 and SimaPro.37 Quasi-optimal values of the utilities; i.e., cooling water, electricity and steam needed for distillation and pervaporation, and travelled distance by the truck are key data input to estimate life cycle inventories. For the sake of comprehension, Tables 21.2.12 and 13 reproduce life cycle inventories obtained in Savelski et al.5 for the production and incineration of 1 kg of isopropyl alcohol, respectively. For the case study under consideration, SimaPro37 was used to calculate most of the life cycle inventories with the exception of the inventories corresponding to the incineration and biological disposal tasks, which were calculated with the aid of Ecosolvent.15 Once the life cycle inventories are estimated, the ReCiPe 200836 method can be used for estimating the environmental burdens of the alternatives considered in terms of the human health, ecosystem, and natural resources as will be shown in the Results section. For more details, the interested reader is referred to the papers of Meyer et al.11 and Amelio et al.16

1692

21.2 Recovery Versus Incineration

Table 21.2.12. Life cycle inventory for IPA manufacture on a 1 kg basis. [Data from reference 5]. Cumulative energy demand [MJ-Eq]

6.26E+01

Air emissions [kg]

1.66E+00

CO2 [kg]

1.63E+00

CO [kg]

2.28E-03

CH4 [kg]

1.03E-02

NOx [kg]

3.04E-03

NMVOC [kg]

1.78E-03

Particulates [kg]

1.27E-03

SO2 [kg]

4.91E-03

Water emissions [kg]

5.80E-01

VOCs [kg]

9.01E-07

Soil emissions [kg]

2.44E-04

Total emissions [kg]

2.24E+00

Table 21.2.13. Life cycle inventory for the incineration of 1 kg of IPA. [Data from reference 5].

CO2 [kg]

2.20E+00

CO [kg]

2.26E-05

NOx [kg]

2.59E-04

NMVOC [kg]

3.09E-06

Particulates [kg]

3.78E-05

Total emissions [kg]

2.21E+00

21.2.6 RESULTS 21.2.6.1 Optimization of Alternative A1: Distillation/incineration As mentioned in Section 21.2.2, the main optimization variable of this alternative is the recovery of isopropyl alcohol in the first cut. The conceptual model for the distillation task was repeatedly solved until achieving the corresponding pseudo-steady state for “perpass” recoveries of IPA in the interval [84-98%]. In all cases, the concentration of IPA in the first cut was set to 83 wt% (Table 21.2.1). Equation 21.2.5 was iteratively solved in each case for the vapor flow rate V at the given value of the distillation processing time tD of 20 h. Figure 21.2.11 graphically depicts the behavior of the vapor flow rate with the optimization variable. The minimum value for V corresponds to a “per-pass” recovery of IPA in the first cut of around 96%. While Table 21.2.14 shows the design and operation variables obtained at the conceptual model level, Table 21.2.15 reports the corresponding investment and operation costs calculated from the cost model on a yearly basis. Among the operating costs, the

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1693

costs involved in transporting the concentrated distillate to the cement plant were estimated from parameters given in Table 21.2.5. Twenty-nine trips with a total traveled distance of 12760 km are required to cover the distance from the pharmaceutical plant to the cement kiln. Parameters needed to estimate the costs associated with the incineration of the distillate and with the biological disposal of the column residue are given in Table 21.2.6.

Figure 21.2.11. Vapor flow rate as a function of the “per-pass” recovery of IPA in the first cut. In all cases, tD = 20 h and pseudo-steady state is achieved. [Adapted, by permission, from R. Meyer, D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual model-based optimization and environmental evaluation of waste solvent technologies: Distillation/incineration versus distillation/pervaporation, Sep. Purif. Technol., 158, 238 (2016)].

Table 21.2.14. Design and operation variables of the batch rectifier for different values of the “per-pass” recovery of IPA in the first cut. AlternativeA1: Distillation/incineration. [Adapted, by permission, from R. Meyer, D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual model-based optimization and environmental evaluation of waste solvent technologies: Distillation/incineration versus distillation/pervaporation, Sep. Purif. Technol., 158, 238 (2016)]. IPA “per-pass” recovery [%] 84

88

92

96

98

2

Condenser area [m ]

1.839

1.775

1.744

1.721

1.826

2

1.402

1.353

1.329

1.312

1.392

7.9

7.7

7.5

7.3

7.2

Evaporator area [m ] Horizontal vessel volume [m3] Packed height [m] Packed diameter [m]

5.0

5.0

5.0

5.0

5.0

0.3048

0.3048

0.3048

0.3048

0.3048

Cooling water [m3/year]

15213

14682

14424

14238

15105

Steam [kg/year]

578661

558486

548671

541583

574571

1694

21.2 Recovery Versus Incineration

Table 21.2.15. Annualized investment and operating costs [U$S/year] for different values of the “per-pass” recovery of IPA in the first cut. Alternative A1: Distillation/incineration. [Adapted, by permission, from R. Meyer, D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual model-based optimization and environmental evaluation of waste solvent technologies: Distillation/incineration versus distillation/ pervaporation, Sep. Purif. Technol., 158, 238 (2016)]. IPA “per-pass” recovery [%] 84

88

92

96

98

Column investment cost

64897

64613

64348

64084

64030

Column utilities cost

16079

15519

15247

15050

15966

Column residue disposal cost

19186

19186

19186

19186

19186

Transportation cost

10756

10756

10756

10756

10756

Incineration cost in cement kiln

85500

85500

85500

85500

85500

General cost

102580

102420

102270

102130

102100

Overall cost

298998

297994

297307

296706

297538

473.8

473.8

473.8

473.8

473.8

631

629

627.5

626.2

628

IPA to incineration [t/year] Overall cost [U$S/t IPA]

Consistent with the results shown in Figure 21.2.11, the optimal recovery of isopropyl alcohol per batch is around 96%. However, as it is shown in Table 21.2.15, any of the design alternatives can be implemented since the differences in annualized costs are rather small. Note that for a batch time of 20 hours, the value that is large enough for the separation of the given feed amount, the rectifier processing capacity measured as its cross-sectional area could accommodate vapor flow rates as high as the double of the actual values. In other words, the separation could be accomplished in less time maintaining the same column. The cost of 296700 U$S/year is equivalent to a cost of 626 U$S/t IPA. Table 21.2.16 shows the mass balance in the pseudo-steady state, achieved after three days, corresponding to the optimal design. The feed charge is 5074 kg, and the amount of isopropyl alcohol sent to incineration is about 97.5% of the solvent contained in the fresh feed. Table 21.2.16. Mass balance in the pseudo-steady state for the optimal design. Alternative A1: Distillation/incineration. [Adapted, by permission, from R. Meyer, D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual model-based optimization and environmental evaluation of waste solvent technologies: Distillation/incineration versus distillation/pervaporation, Sep. Purif. Technol., 158, 238 (2016)].

Overall amount [kmol]

Column charge

Cut1

Residue1

Cut2 (Recycle)

Residue2

228.28

36.57

191.71

3.26

188.45

IPA [mol/mol]

0.1001

0.6000

0.0048

0.1077

0.0030

Water [mol/mol]

0.8999

0.4000

0.9952

0.8923

0.9970

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1695

21.2.6.2 Optimization of Alternative A2: Distillation/pervaporation According to the scheduling proposed in Figure 21.2.2, different separations were simulated with operation times for distillation and dehydration by pervaporation of 16 and 20 hours, respectively. The “per-pass” recovery of isopropyl alcohol in the batch rectifier was set at 92%, and variants with an IPA mole fraction in the first cut in the range [0.6-0.675] were analyzed. The composition of the distillate was selected as the main optimization variable to properly capture the cost trade-offs between the distillation column and the membrane unit. Regarding the distillation task, Figure 21.2.5 depicts the time evolution of the primary variables calculated at the conceptual model level for an IPA mole fraction in the first cut of 0.65; that in turn is the optimal value from an economic point of view. The vapor flow rate V necessary to accomplish the desired separation in 16 hours is 5.359 kmol/h. The required number of theoretical stages and diameter of the rectifier are 13 and 0.3048 m, respectively. Table 21.2.17 shows the mass balances in the pseudo-steady state corresponding to this variant. Note that the charge to the column boiler (5568 kg) is composed by the fresh feed, the second cut and the condensed permeate from the membrane unit. Table 21.2.18 shows the corresponding mass balances in the pseudo-steady state for the IPA dehydration step in the membrane unit. The amount of isopropyl alcohol recovered in the product tank represents an overall recovery of 97.3% of the solvent contained in the fresh feed. Table 21.2.17. Mass balance in the pseudo-steady state for the optimal design corresponding to the distillation step. Alternative 2: Distillation/pervaporation. [Adapted, by permission, from R. Meyer, D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual model-based optimization and environmental evaluation of waste solvent technologies: Distillation/incineration versus distillation/pervaporation, Sep. Purif. Technol., 158, 238 (2016)].

Column charge

Cut1

Residue1

Cut2 (Recycle)

Residue2

Overall amount [kmol]

248.77

36.54

212.22

9.34

202.88

IPA [mol/mol]

0.1038

0.6500

0.0097

0.1543

0.0030

Water [mol/mol]

0.8962

0.3500

0.9903

0.8457

0.9970

Table 21.2.18. Mass balance in the pseudo-steady state for the optimal design corresponding to the purification step in the membrane unit. Alternative 2: Distillation/pervaporation. [Adapted, by permission, from R. Meyer, D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual model-based optimization and environmental evaluation of waste solvent technologies: Distillation/incineration versus distillation/ pervaporation, Sep. Purif. Technol., 158, 238 (2016)]. Retentate vessel charge = Cut1

Product

Permeate (Recycle)

Overall amount [kmol]

36.54

22.00

14.54

IPA [mol/mol]

0.6500

0.9949

0.1281

Water [mol/mol]

0.3500

0.0051

0.8719

1696

21.2 Recovery Versus Incineration

Figure 21.2.12 shows the simulation results of the purification step in the membrane unit corresponding to the feed charge reported in Table 21.2.18. Simulations in gPROMS31 were done for different values of the membrane area until achieving an IPA purity equal to or greater than 99.4 mole percent (99.8 wt%) in a pervaporation processing time equal or lower than 20 hours. For the optimum variant the required area was 87.5 m2. Modules of 17.5 m2 each working in parallel were considered in order to maximize the driving force through the membrane along the batch time. Figure 21.2.8 in Section 21.2.3.2 shows the evolution of the temperature Figure 21.2.12. Evolution of the mole composition of isopropyl alcohol in the product tank corresponding to along the retentate side of the membrane the optimal design. Alternative 2: Distillation/pervapounit corresponding to the optimal design. ration. [Adapted, by permission, from R. Meyer, D.A. Tables 21.2.19 and 20 show the results Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual obtained at the conceptual model level and model-based optimization and environmental evaluation of waste solvent technologies: Distillation/inciner- the cost model level, respectively. As mentioned above, the optimal IPA mole fraction ation versus distillation/pervaporation, Sep. Purif. Technol., 158, 238 (2016)]. in the first cut is about 0.65. At higher values there is a substantial increase in operating and investment costs of the column that does not outweigh the savings in membrane area caused by feeds with higher alcohol composition and lower amounts to be processed with respect to the rest of the designs studied. The cost of recovery for the optimal composition amounts to 856 U$S/t IPA, below the current purchase cost of IPA of 1200 U$S/t IPA reported in Table 21.2.6. Table 21.2.19. Design and operation variables for different values of the IPA mole fraction in the first cut. Alternative A2: Distillation/pervaporation. [Adapted, by permission, from R. Meyer, D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual model-based optimization and environmental evaluation of waste solvent technologies: Distillation/incineration versus distillation/pervaporation, Sep. Purif. Technol., 158, 238 (2016)]. IPA mole fraction 0.6

0.625

0.65

0.675

Column 2

Condenser area [m ]

2.351

2.355

1.693

1.645

Evaporator area [m2]

1.792

1.795

1.290

1.254

Horizontal vessel volume [m3]

7.36

7.36

7.36

7.36

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1697

Table 21.2.19. Design and operation variables for different values of the IPA mole fraction in the first cut. Alternative A2: Distillation/pervaporation. [Adapted, by permission, from R. Meyer, D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual model-based optimization and environmental evaluation of waste solvent technologies: Distillation/incineration versus distillation/pervaporation, Sep. Purif. Technol., 158, 238 (2016)]. IPA mole fraction 0.6 Packed height [m] Packed diameter [m]

0.625

0.65

0.675

5.0

5.5

8.0

14.0

0.3048

0.3048

0.3048

0.3048

Cooling water [m3/year]

15556

15582

15364

16473

Steam [kg/year]

591720

592701

584413

626616

Membrane Unit Vessel [m3]

3

2.95

2.9

2.86

Vacuum pump duty [kW]

1.60

1.59

1.46

1.44

Compressor duty [kW]

7.75

7.50

6.03

6.80

2

Permeate condenser area [m ]

8.4

8.1

6.5

6.2

Retentate heater area [m2]

1.05

1.01

0.81

0.77

Propane condenser area [m2]

4.4

4.2

3.4

3.2

Membrane area [m2]

105

105

87.5

87.5

Vacuum vessel volume [m3]

5.3

5.3

4.7

4.7

Cooling water [m3/year]

17776

16568

14851

13678

Steam [kg/year]

136577

124887

112790

101700

Energy consumption [kWh/year]

37846

35056

33529

30828

Table 21.2.20. Annualized investment and operating costs [U$S/year] for different values of the IPA mole fraction in the first cut. Alternative 2: Distillation/pervaporation. [Data from reference 11]. IPA mole fraction 0.6

0.625

0.65

0.675

Pervaporation unit investment cost

126678

125938

113874

113139

Column investment cost

64585

66070

72674

87271

Column utilities cost

16442

16470

16239

17412

Column residue disposal cost

19188

19188

19188

19188

Pervaporation unit utilities cost

5084

4677

4214

3826

Pervaporation unit electricity cost

4640

4298

4110

3779

1698

21.2 Recovery Versus Incineration

Table 21.2.20. Annualized investment and operating costs [U$S/year] for different values of the IPA mole fraction in the first cut. Alternative 2: Distillation/pervaporation. [Data from reference 11]. IPA mole fraction 0.6

0.625

0.65

0.675

5308

5308

4423

4423

General cost

173077

173493

170448

178181

Overall cost

415002

415442

405170

427219

Membrane replacement cost

IPA recovered [t/year] Overall recovery cost [U$S/t IPA]

473

473

473

473

877.4

878.3

856.6

903.2

21.2.6.3 Comparison between the treatment alternatives from an economic standpoint Table 21.2.21 summarizes the economic figures obtained at the corresponding optimum for alternatives A1 and A2, respectively. The overall cost of each alternative, taking into account the cost of virgin solvent needed in each case, is also reported. Annual earnings of the alternative A2 with respect to the alternative A1, considering savings due to the avoided purchase of fresh solvent and the avoided disposal cost of the spent solvent, amount 459000 U$S/year. Thus, considering an income tax rate of 35% and a 10% interest rate, a payback period of 2.2 years and an internal return rate of 33% is calculated for the recovery alternative. From these results, it is clear that the recovery alternative is financially advantageous. Table 21.2.21. Overall costs [U$S/t IPA] for alternatives distillation/incineration and distillation/pervaporation at optimal designs including the purchase cost of virgin solvent. [Data from reference 11]. A1: Distillation/incineration Investment cost [U$S/year]

64084

Utilities cost [U$S/year]

34236

Transport cost [U$S/year]

10756

Incineration cost in cement kiln [U$S/year]

85500

General cost [U$S/year]

102130

Overall cost [U$S/year]

296706

Virgin solvent requirement [t IPA/year]

486

Overall cost including the purchase cost of virgin solvent [U$S/t IPA]

1810.5

A2: Distillation/pervaporation Investment cost [U$S/year]

186548

Utilities cost [U$S/year]

48175

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1699

Table 21.2.21. Overall costs [U$S/t IPA] for alternatives distillation/incineration and distillation/pervaporation at optimal designs including the purchase cost of virgin solvent. [Data from reference 11]. General cost [U$S/year]

170448

Overall cost [U$S/year]

405170

Virgin solvent requirement [t IPA/year]

13

Overall cost including the purchase cost of virgin solvent [U$S/t IPA]

865.8

21.2.6.4 Comparison between the treatment alternatives from environmental considerations For alternative A1, Table 21.2.22 shows the input data necessary to estimate the environmental impacts corresponding to i) the production of virgin solvent, ii) the distillation task, iii) the biological disposal of the column residue, iv) the transportation of the distillate from the pharmaceutical plant to the cement kiln and v) the incineration task. Table 21.2.23, on the other hand, summarizes the input data needed to calculate the environmental burden of alternative A2. Data for the following tasks are considered for the distillation/pervaporation alternative: i) virgin IPA production, ii) distillation, iii) pervaporation, iv) biological disposal and v) vacuum production with a vacuum pump/refrigeration system. Table 21.2.22. Input data for LCA. Alternative A1: Distillation/incineration. [Data from reference 11] Overall mass balance per batch, one batch of 20 hours per day, 360 batches/year Overall Amount (kg/batch) Isopropyl alcohol (mass fraction) Water (mass fraction)

Feed 5000 0.27 0.73

Distillate 1582.2 0.8334 0.1666

Energy balance per batch, one batch of 20 hours per day, 360 batches/year Steam at 2 bar (kg/batch) Cooling water (m3/batch) Virgin solvent requirement (t/year)

1504.4 39.5 486

Transportation of the distillate from Pharmaceutical plant to Cement kiln Number of trips/year Distance/trip (km/trip) Fuel economy (km/L) Fuel consumption (L)

29 440 5 2552

Residue 3417.8 0.0092 0.9908

1700

21.2 Recovery Versus Incineration

Table 21.2.23. Input data for LCA. Alternative A2: Distillation/pervaporation. [Data from reference 11]. Overall mass balance per batch, one batch per day, 360 batches/year Overall Amount (kg/batch) Isopropyl alcohol (mass fraction) Water (mass fraction)

Feed 5000 0.27 0.73

Product 1317.4 0.9985 0.0015

Residue 3682.6 0.0094 0.9906

Energy balance per batch, one batch per day, 360 batches/year Distillation Column (18 h/batch) Steam at 2 bar (kg/batch) Cooling water (m3/batch Membrane unit (19.45 h/batch) Steam at 2 bar (kg/batch) Cooling water (m3/batch) Compressor electricity consumption (kWh/batch) Vacuum pump electricity consumption (kWh/batch) Circulating pump electricity consumption (kWh/batch) Virgin solvent requirement (t/year)

1623.4 42.7 313.3 41.3 66.8 25.1 1.2 13

Figure 21.2.13 shows the environmental impact caused by the two treatment technologies considered at the endpoint level in category indicators damage to human health, damage to ecosystem diversity and damage to resource availability. The environmental burden associated with variant A1-WSI (i.e., distillation followed by off-site incineration in a conventional waste solvent incinerator) is also reported. The results show that alter36 Figure 21.2.13. Environmental impact (ReCiPe 2008) caused in native A1-WSI generates the most endpoints categories human health, ecosystem and resources by alternatives A1 and A2. [Data from reference 11]. significant impact. Alternative A1cement kiln shows a burden on the availability of resources while bringing benefits for endpoints human health and ecosystem quality. On the other hand, alternative A2 does not show any benefit. However, its burden, in particular on resources availability, is rather small whether it is compared with those corresponding to both incineration alternatives mainly due to the almost complete recovery of IPA using the hybrid process distillation/pervaporation. For this alternative, the burden corresponding to the production of IPA is calculated from a virgin solvent requirement of 13 t IPA/year while burdens for the incineration variants must be estimated

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa

1701

from an IPA requirement amounting 486 t IPA/year. According to Table 21.2.12, a significant reduction in the cumulative energy demand is observed for the production of IPA in alternative A2 with respect to those in alternatives A1, as it is reduced in 29.6 E06 MJ/ year. For alternative A1-cement kiln, the burden on midpoint fossil fuel depletion is lower than that of alternative A1-WSI because in conventional incinerators additional energy is needed to burn the solvent. Moreover, a definite difference in results in favor of alternative A1-cement kiln can also be found in the category indicators at midpoint level climate change, human toxicity, and particulate matter formation. In line with these results, a positive effect on the environment is expected for this alternative because heavy fuel, hard coal, and their corresponding emissions are avoided. Regarding the decision-making process, alternative A2 is the best choice from an environmental perspective if the designer is interested in minimizing the use of resources even when the hybrid process distillation/pervaporation produces a small impact regarding human health and ecosystem quality. On the other hand, when ecosystem quality and, in particular, human health are paramount, alternative A1 with incineration in a cement kiln is the best option. For the case study under consideration, we emphasize the importance of using and developing solvent recovery technologies to minimize the impact associated with solvent manufacture as noticed in several recent works in this field.1,2,5,11,16 21.2.7 DISCUSSION Focusing the attention on the mixture isopropyl alcohol-water, quasi-optimal designs for the alternatives distillation/incineration and distillation/pervaporation were obtained from the conceptual design of the two waste solvent technologies. The main advantage of the conceptual design approach is that both process trade-offs and thermodynamic constraints of the mixture can be properly captured at a low computational cost. Indeed, pseudosteady state solutions, which are normally neglected when the process is modeled by rigorous simulation, are easily handled in the conceptual modeling framework. Optimal values for “per-pass” recovery of isopropyl alcohol and distillate composition were determined for the first and second alternative, respectively. From the analysis of the economic figures, it was concluded that solvent recovery through the hybrid process is the best option mainly due to the savings in the purchase of virgin solvent. The economic analysis was complemented by a life cycle assessment, with input data from the conceptual design step, as a tool to determine which technology is the most appropriate for the treatment of the waste solvent from an environmental standpoint. From the life cycle analysis, it was observed that the main impact on the environment was caused by the solvent production. The environmental impact caused by solvent recovery showed the lowest values regarding a total score indicating that the hybrid process the best option if minimization of the use of resources is paramount. Regarding the performance assessment approach, it was concluded that life cycle analysis is enriched by input data obtained from the conceptual model-based process optimization of each alternative studied, especially regarding process knowledge and reliability of results.

1702

21.2 Recovery Versus Incineration

However, a step forward must be taken in the proposed approach to incorporate in the analysis the recovery of different low-volume waste streams generated from multiple production campaigns with the same recovery system. Also, not only sharp but also sloppy splits considering partial recovery of solvents and, consequently, the incineration of the non-recovered fractions must be taken into account. In this way, implementation of the recovery system will become economically feasible while reducing the environmental impact of the pharmaceutical process. Table 21.2.24 summarizes the encouraging results recently obtained in this field by Savelski et al.5 Table 21.2.24 Economic evaluation of solvent recovery in selamectin, nelfinavir, and hydrocortisone processes. [Adapted, by permission, from M.J. Savelski, C.S. Slater, P.V Tozzi, and C.M. Wisniewski, On the simulation, economic analysis, and life cycle assessment of batch-mode organic solvent recovery alternatives for the pharmaceutical industry, Clean Techn. Environ. Policy, 19 (10), 2467 (2017)].

Recovery scenario

Selamectin (alone) Nelfinavir (alone) Hydrocortisone (alone) Selamectin + Nelfinavir Selamectin + Hydrocortisone Selamectin + Nelfinavir + Hydrocortisone

Operating cost savings with respect to the current process using incineration as a mean of disposal (U$S/year)*

Payback period (year)

Internal return rate (%)

215000 80000 252000 314000 486000

8.8 n/a 7.9 6.9 5.1

8.7 -1.3 11 14.5 23.2

584000

4.5

27.9

* Annual maintenance cost included in operating cost savings

21.2.8 ACKNOWLEDGEMENTS José Espinosa dedicates this chapter to therapists and personnel of the Research and Rehabilitation Center “Doctor Esteban Laureano Maradona" for their outstanding work during his rehabilitation therapy from a cerebrovascular accident. REFERENCES 1 2 3 4

P. Luis, A. Amelio, S. Vreysen, V. Calabro, and B. Van der Bruggen, Life cycle assessment of alternatives for waste-solvent valorization: batch and continuous distillation vs incineration, Int. J. Life Cycle Assess., 18, 1048 (2013). A. Amelio, G. Giuseppe, S. Vreysen, P. Luis, and B. Van der Bruggen, Guidelines based on life cycle assessment for solvent selection during the process design and evaluation of treatment alternatives, Green Chem., 16, 3045 (2014). C. Seyler, S. Hellweg, M. Monteil, and K. Hungerbühler, Life cycle inventory for use of waste solvent as fuel substitute in the cement industry: a multi-input allocation model, Int. J. Life Cycle Assess., 10 (2), 120 (2004). C. Seyler, T.B. Hofstetter, and K. Hungerbühler, Life cycle inventory for thermal treatment of waste solvent from chemical industry: a multi-input allocation model, J. Clean Prod., 13, 1211 (2005).

*Danilo Alexander Figueroa Paredes, **Antonio Amelio, *José Espinosa 5 6 7 8 9 10 11

12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

1703

M.J. Savelski, C.S. Slater, P.V Tozzi, and C.M. Wisniewski, On the simulation, economic analysis, and life cycle assessment of batch-mode organic solvent recovery alternatives for the pharmaceutical industry, Clean Techn. Environ. Policy, 19 (10), 2467 (2017). I.M. Smallwood, Solvent Recovery Handbook, Second edition, CRC Press, Boca Raton, 2002. A.M. Urtiaga, E.D. Gorri, and I. Ortiz, Pervaporative recovery of isopropanol from industrial effluents, Sep. Purif. Technol., 49, 245 (2006). K. Koczka, J. Manczinger, P. Mizsey, and Z. Fonyo, Novel hybrid separation processes based on pervaporation for THF recovery, Chem. Eng. Process., 46, 239 (2007). C.S. Slater, M.J. Savelski, T.M. Moroz, and M.J. Raymond, Pervaporation as a green drying process for tetrahydrofuran recovery in pharmaceutical synthesis, Green Chemistry Letters and Reviews, 5 (1), 55 (2012). P. Luis, A. Amelio, S. Vreysen, V. Calabro, and B. Van der Bruggen, Simulation and environmental evaluation of process design: Distillation vs. hybrid distillation–pervaporation for methanol/ tetrahydrofuran separation, Applied Energy, 113, 565 (2014). R. Meyer, D.A. Figueroa Paredes, M. Fuentes, A. Amelio, B. Morero, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual model-based optimization and environmental evaluation of waste solvent technologies: Distillation/incineration versus distillation/pervaporation, Sep. Purif. Technol., 158, 238 (2016). M. Skiborowski, A. Harwardt, and W. Marquardt, Conceptual design of distillation based hybrid separation processes, Annu. Rev. Chem. Biomol. Eng., 4, 45 (2013). M.A. Sosa, D.A. Figueroa Paredes, J.C. Basílico, B. van der Bruggen, and J. Espinosa, Screening of pervaporation membranes with the aid of Conceptual Models: an application to bioethanol production, Sep. Purif. Technol., 146, 326 (2015). T.B. Hofstetter, C. Capello, and K. Hungerbühler, Environmentally preferable treatment options for industrial waste solvent management, a case study of a toluene containing waste solvent, TransIchemE, 81 (Part B), 189 (2003). C. Capello, S. Hellweg, B. Badertscher, H. Betschart, and K. Hungerbühler, Environmental assessment of waste-solvent treatment options. Part I: The Ecosolvent tool, J. Ind. Ecol., 11 (4), 26 (2007). A. Amelio, D.A. Figueroa Paredes, J. Degrève, P. Luis, B. Van der Bruggen, and J. Espinosa, Conceptual Model-Based Design and Environmental Evaluation of Waste Solvent Technologies: Application to the separation of the mixture acetone-water, Sep. Sci. Technol., 53 (11), 1791, (2018). W. Marquardt, S. Kossack, and K. Kraemer, A Framework for the Systematic Design of Hybrid Separation Processes, Chinese J. Chem. Eng., 16, 333 (2008). J.A. Caballero, I.E. Grossmann, M. Keyvani, and E.S. Lenz, Design of hybrid distillation-vapor membrane separation systems, Ind. Eng. Chem. Res., 48, 9151 (2009). K. Koch, D. Sudhoff, S. Kreiß, A. Górak, and P. Kreis, Optimisation-based design method for membrane-assisted separation processes, Chem. Eng. Process. Process Intensif., 67, 2 (2013). M. Skiborowski, J. Wessel, and W. Marquardt, Efficient Optimization-Based Design of Membrane-Assisted Distillation Processes, Ind. Eng. Chem. Res., 53, 15698 (2014). J. Espinosa, and J.L. Marchetti, Conceptual modeling and referential control applied to batch distillations, Ind. Eng. Chem. Res., 46, 6000 (2007). K.A. Torres, and J. Espinosa, The influence of tangent pinch points on the energy demand of batch distillations: development of a conceptual model for binary mixtures, Ind. Eng. Chem. Res., 50, 6260 (2011). M.F. Doherty, and M.F. Malone, Conceptual Design of Distillation Systems, McGraw-Hill Higher Education, New York, 2001. C. Bernot, M.F. Doherty, and M. F.; Malone, Design and operating targets for nonideal multicomponent batch distillation, Ind. Eng. Chem. Res., 32, 293 (1993). Distil User Manual, Hyprotech Ltd., Calgary, Canada, 1999. Aspen Batch Distillation & Aspen Hysys User Manual, Version 8, 2014. I.M Mujtaba, and S. Macchietto, Holdup issues in batch distillation-binary mixtures, Chem. Eng. Sci., 53 (14), 2519 (1998). G. Genduso, P. Luis, and B. Van der Bruggen, Overcoming any configuration limitation: an alternative operating mode for pervaporation and vapour permeation, J. Chem. Technol. Biotechnol., (2015) T. Melin, and R. Rautenbach, Membranverfahren: Grundlagen der Modul- und Anlagenauslegung, 2nd ed., Springer-Verlag, Berlin, Heidelberg, New York, 2004. M.A. Sosa, and J. Espinosa, Feasibility Analysis of Isopropanol Recovery by Hybrid Distillation/ Pervaporation Process with the Aid of Conceptual Models, Sep. Purif. Technol., 78, 237 (2011).

1704 31 32 33 34 35 36 37 38

21.2 Recovery Versus Incineration

gPROMS User Manual, Process Systems Enterprise Ltd., 1998. A. Fahmy, Membrane Processes for the Dehydration of Organic Compounds, Ph. D. Thesis, University of Hannover, Germany (2002). G.L. Ding, Recent developments in simulation techniques for vapour-compression refrigeration systems. Int. J. Refrig., 30, 1119 (2007). A.K. Coker, Ejectors and Mechanical Vacuum Systems, Ludwig’s Appl. Process Des. Chem. Petrochemical Plants, 4th ed., Elsevier, Burlington, 2007. W.D. Seider, J.D. Seader, and D.R. Lewin, Product and Process Design Principles: Synthesis, Analysis, and Evaluation, 2nd intern, John Wiley & Sons, New York, 2004. M. Goedkoop, R. Heijungs, M. Huijbregts, A. De Schryver, J. Struijs, and R. Van Zelm, ReCiPe 2008: a life cycle impact assessment method which comprises harmonised category indicators at the midpoint and the endpoint level. Report I: Characterisation (2009). SimaPro User Manual, Version 7.3, PréConsultants, 2001. A.W. Sleeswijk, L.F. van Oers, J.B. Guinée, J. Struijs, and M.A. Huijbregts, Normalisation in product life cycle assessment: An LCA of the global and European economic systems in the year 2000. Sci. Total Environ., 390, 227 (2008).

George Wypych

1705

21.3 SOLVENT RECOVERY, RECYCLING, AND INCINERATION George Wypych ChemTec Laboratories, Toronto, Canada Distillation is a millennia old technology of solvent recovery used for various compositions of materials, including modern chemical processes. To take full advantage of natural resources distillation has to be carefully analyzed considering energy input required and costs and benefits of recovery. Optimal design of advanced distillation configuration for enhanced energy efficiency of waste solvent recovery process in semiconductor industry is one of the examples of developments.1 A distillation process was used to recover waste solvent, reduce the production-related costs, and protect environment from the semiconductor industrial waste.1 The conventional sequence of distillation columns was optimized using the Box and sequential quadratic programming method for the minimum energy use.1 A heat pump-assisted distillation was implemented to improve the distillation performance achieving a significant energy savings of 40.5% compared to the conventional column sequence.1 Figure 21.3.1 shows a complexity of the system.1 The thermally coupling of columns 3 and 5 helped to enhance energy efficiency further improved by application of heat integration and heat pump to column 4.1 The distillation technique was used to recover monoisopropanolamine, methyl diglycol, N-methylformamide, and 1-piperazine ethanol from the wastes of a photoresist stripper in the thin-film transistor light emitting diodes industry.1

Figure 21.3.1. Solvent recovery from semiconductor wastes using heat integration and heat pump assisted distillation. [Adapted, by permission, from Y. D. Chaniago, L. Q. Minh, M. S. Khan, K.-K. Koo, A. Bahadori, M. Lee, Ener. Conversion Manag., 102, 92-103, 2015.]

1706

21.3 Solvent Recovery, Recycling, and Incineration

Figure 21.3.2. Aspen Hysys™ flow sheet for the Rich Vapor Compression (RVC) process configuration. [Adapted, by permission, from L. Dubois, D. Thomas, Int. J. Greenhouse Gas Control, 69, 20-35, 2018.]

Carbon dioxide absorption is an integral part of many processes, such as natural gas or biogas to achieve heating value requirements, to prevent freezing in the low-temperature chillers in the production of liquefied natural gas, to avoid catalyst poisoning in the manufacture of ammonia, decrease global warming potential caused by carbon dioxide emissions, and obtaining supercritical liquid carbon dioxide.2 Chemical absorption columns are most frequently implemented for this purpose.2 Ammonia, monoethanolamine, diethanolamine, triethanolamine, 2-aminoethoxyethanol, diisopropanolamine, diethanolmethylamine use and recovery have been studied.2 The higher the concentration of CO2 absorbed, the lower the amount of solvent required.2 The carbon dioxide diffusion in the aqueous solvent is related to the size of the absorption column required.2 The post-combustion CO2 capture process in cement producing plants by absorptionregeneration is the more mature technology but its cost reduction is still required.3 Three different solvents, namely monoethanolamine, piperazine, and piperazine-methyldiethanolamine blend have been studied. The heat pump modifications, namely LVC and RVC lead to the higher regeneration energy savings (from 11% to 18% depending on the solvent) while also reducing significantly the condenser cooling energy.3 The best results in terms of the regeneration energy were obtained with the blend of 10 wt% monoethanolamine + 30 wt% piperazine using RVC configuration (Figure 21.3.2).3 A solvent for recovery of carbon dioxide from gaseous mixture contains alkanolamine, reactive amines acting as promoter or activators, sulfolane, and a carbonate buffer.4 Success of electric vehicles requires reduction of the cost of the lithium-ion battery. Drying the coated cathode layer and subsequent recovery of the solvent for recycle is a vital step in the lithium-ion battery manufacturing.5 The energy demand for the process is very significant at 10.2 kWh per kg of n-methyl pyrrolidone vaporized, and 421 kWh per 10 kWh battery pack.5 The overall energy demand is 45 times of the heat needed to vapor-

George Wypych

1707

ize the n-methyl pyrrolidone.5 The large energy demand is a result of the large quantity of air that has to be heated and cooled.5 The cost of solvent recovery consists 3.4% of the total cost of production of battery.5 The solvent deasphalting process is a heavy oil upgrading process in which deasphalted oil is extracted from heavy oil feedstock by precipitating asphaltene using an excess amount of alkane solvent.6 After the extraction, solvent recovery is carried out to separate the solvent from the deasphalted oil to recycle the expensive solvent.6 CarFigure 21.3.3. Solvent recovery improvement. bon dioxide was applied to selectively sepa[Adapted, by permission, from S. I. Im, S. Shin, J. W. Park, H. J. Yoon, K. S. Go, N. S. Nho, K B. Lee, Chem. rate solvent from deasphalted oil at a Eng. J., 331, 389-94, 2018.] relatively low temperature.6 The temperature was decreased from 200°C to 40°C when using CO2 compared to the conventional method.5 The increase in the interaction between CO2 and solvent causes the separation of solvent from deasphalted oil, leading to an increase in solvent recovery as can be deduced from verification based on Hansen distance calculations (Figure 21.4.3).6 The numerical simulations indicate the possibility of continuous operation of solvent recovery using carbon dioxide.6 Oils and fats contain wide variety of chemical building blocks, including fatty acids, fatty acid esters, and fatty alcohols which can be used as a feedstock for the production of renewable materials.7 An efficient recovery of non-polar extraction solvents is possible for a wide range of low-quality oils.7 Commercially available PDMS membranes enable high solvent fluxes and rejection of triglyceride and fatty acids.7 In deacidification experiments, a highly-selective separation of free fatty acids from triglycerides for long-chain and medium-chain oils can be achieved using polar solvents and available polyimidebased membranes.7 A membrane performance can be accurately described by a (reduced) solution-diffusion model with imperfections, for which mass transfer of solvents and fatty acids is described by the convective contribution, while mass transfer of triglycerides is described by the diffusive contribution.7 A membrane vapor extraction efficiently recovers butanol from a dilute aqueous solution, such as, for example, from a fermentation broth.8 In the membrane vapor extraction, feed and solvent liquids are not in contact; they are separated by vapor.8 A semi-volatile aqueous solute, such as butanol vaporizes at the upstream side of a membrane, diffuses as a vapor through the membrane pores, and subsequently condenses and dissolves into a high-boiling nonpolar solvent, favorable to the solute but not to water.8 Compared to the traditional distillation, the energy consumption with the membrane vapor extraction is minimal owing to essential cancellation of the enthalpies of feed vaporization and absorption by solvent.8 Because transport across the membrane occurs in the vapor

1708

21.3 Solvent Recovery, Recycling, and Incineration

phase rather than in the liquid phase, membrane-transport resistance is significantly reduced.8 A three-stage organic solvent nanofiltration process has been presented which comprises of a two-stage membrane cascade for separation with a third membrane stage added for integrated solvent recovery, i.e., solvent recycling.9 The two-stage cascade gives an increased separation selectivity whereas the integrated solvent recovery stage mitigates the otherwise large solvent consumption of the purification.9 This operation attains a purity of 98.7% through semi-continuous process with two washes of the solvent recovery stage.9 Extraction of vegetable oils results in a mixture of 70-75 wt% hexane and 25-30 wt% oil.10 The organic solvent nanofiltration can be used to recover hexane from the crude mixture of soybean oil and hexane.10 The fluoropolymer-based membranes are promising candidates in aliphatic solvent separation which permit oil rejection above 98%.10 The recycling and recovery of ionic liquids after the extraction of metal ions were investigated for trioctylmethyl ammonium camphorate and trioctylmethyl ammonium dodecanedioate.11 The metal ions extracted were cadmium, copper, and lead with initial concentrations of 50 and 100 ppm.11 The extraction percentage was above 90%.11 The ionic liquid used was water immiscible.11 The ionic liquids were successfully recovered by using solvents with polarity indices between 5.1 and 4.3 (ethanol, acetone).11 The possibility of reusing an ionic liquid for the pretreatment of biomass has been investigated.12 The presence of lignin and water within a pretreatment solvent resulted in a far less effective pretreatment process.12 Ethyl-3-methylimidazolium aceFigure 21.3.4. Illustration of the mechanism of action of tate/ethanolamine (60/40 vol%) provided a switchable solvent. [Adapted, by permission, from Q. multiuse process with deterioration of the Chen, L. Wang, G. Ren, Q. Liu, Z. Xu, D. Sun, J. Col- ionic liquid observed from considerably loid Interface Sci., 504, 645-51, 2017.] low sugar conversions and lignin extraction after using the 5th and 7th batch, respectively.12 Ion-pair interactions were explored to design a fatty acid solvent of switchable miscibility with water.13 Fatty acids of medium-length chains are immiscible with water but become miscible when ion pairs are formed with amines.13 The ion pairs become phase separated after bubbling CO2 into the solution due to the dissociation of the fatty acidamine complexes.13 The switchable solvent mechanism was used in removing and recovering oily contaminants. Figure 21.3.4 illustrates the mechanism of action of a switchable solvent.13 Chlorinated volatile organic compounds, including polychloromethanes, polychloroethanes and polychloroethylenes, are or were used as solvents, degreasing agents in a variety of commercial products.14 These compounds are ubiquitous contaminants that can be found in the contaminated soil, air, and any kind of fluvial medium such as groundwater, river, and lake.14 The review article discusses various remediation technologies including thermal incineration, biodegradation, phytoremediation, ozonation, photocatalytic

George Wypych

1709

oxidation, Fenton oxidation, electrochemical oxidation, zero valent metal and bimetal reduction, catalytic dehydrochlorination, and electrochemical reduction.14 Waste solvents are frequently incinerated for a partial recovery of their combustion heat.15 The large chemical sites often have in-house waste incineration facilities but smaller enterprises outsource the treatment of their residuals which requires the special transfer of information about the contents of wastes.15 A waste classification system developed in Switzerland is based on a heating value and on both water and pollutant content of the residues.15 The generated knowledge is a crucial step in the design and operation of sustainable industrial waste incineration network, aimed at maximizing the energy efficiency while fulfilling technical and regulatory constraints.15 The substitution of combustion fuels in cement plants is increasing throughout many countries, and its individual performance is constantly assessed against strict regulatory standards.16 The waste co-incineration reduced health-critical congeners of dioxins and dioxin-like polychlorinated biphenyls whilst providing the necessary energy and calcination needs.16 REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Y. D. Chaniago, L. Q. Minh, M. S. Khan, K.-K. Koo, A. Bahadori, M. Lee, Ener. Conversion Manag., 102, 92-103, 2015. C. Mendez-Alvarez, V. Plesu, A. E. Bonet Ruiz, J. Bonet Ruiz, P. Iancu, J. Llorens, Proceedings of the 26th European Symposium on Computer Aided Process Engineering – ESCAPE 26, Portorož, Slovenia. L. Dubois, D. Thomas, Int. J. Greenhouse Gas Control, 69, 20-35, 2018. P. B. R. K. Ausula, A. Sharma, WO2014118633A3, Carbon Clean Solutions Pvt, Ltd, Feb. 19, 2015. S. Ahmed, P. A. Nelson, K. G. Gallagher, D. W. Dees, J. Power Sources, 322, 169-78, 2016. S. I. Im, S. Shin, J. W. Park, H. J. Yoon, K. S. Go, N. S. Nho, K B. Lee, Chem. Eng. J., 331, 389-94, 2018. K. Werth, P. Kaupenjohann, M. Knierbein, M. Skiborowski, J. Membrane Sci., 528, 369-80, 2017. J. Chen, N. Razdan, T. Field, D. E. Liu, P. Wolski, X. Cao, J. M. Prausnitz, C. J. Radke, J. Membrane Sci., 528, 103-11, 2017. M. Schaepertoens, C. Didaskalou, J. F. Kim, A. G. Livingston, G. Szekely, J. Membrane Sci., 514, 646-58, 2016. X. Li, W. Cai, T. Wang, Z. Wu, J. Wang, X. He, J. Li, Separation Purification Technol., 181, 223-9, 2017. M. A. Valdes Vergara, I. V. Lijanova, N. V. Likhanova, O. O. Xometl, D. Jaramillo Vigueras, A. J. Morales Ramirez, Separation Purification Technol., 155, 110-7, 2015. P. Weerachanchai, J.-M. Lee, Bioresource Technol., 169, 336-43, 2014. Q. Chen, L. Wang, G. Ren, Q. Liu, Z. Xu, D. Sun, J. Colloid Interface Sci., 504, 645-51, 2017. B. Huang, C. Lei, C. Wei, G. Zeng, Environ. Int., 71, 118-38, 2014. V. Bolis, E. Capon-García, O. Weder, K. Hungerbühler, J. Cleaner Prod., 183, 1228-40, 2018. G. Richards, I. E. Agranovski, J. Hazard. Mater., 323, 698-709, 2017.

1710

21.4 Application of Solar Photocatalytic Oxidation to VOC-Containing Airstreams

21.4 APPLICATION OF SOLAR PHOTOCATALYTIC OXIDATION TO VOC-CONTAINING AIRSTREAMS K. A. Magrini, A. S. Watt, L. C. Boyd, E. J. Wolfrum, S. A. Larson, C. Roth, G. C. Glatzmaier National Renewable Energy Laboratory, Golden, CO, USA

21.4.1 SOLVENT DEGRADATION BY PHOTOCATALYTIC OXIDATION Photocatalytic solvent oxidation had been demonstrated at the pilot scale in two field tests located at McClellan Air Force Base (AFB) in Sacramento, California and at the Fort Carson U. S. Army Installation in Colorado Springs, Colorado (Watt et al., 1999; Magrini et al., 1998). The objective of the tests was to determine the effectiveness of solar-powered photocatalytic oxidation (PCO) treatment units for destroying emissions of chlorinated organic compounds (trichloroethylene and dichloroethylenes) from an air stripper at ambient temperature and destroying paint solvent emissions (toluene and MEK) from a painting facility at higher temperatures. Goals for field testing these solar-driven systems were to gather real-world treatability data and establish that the systems maintained performance during the duration of the testing. The photocatalytic oxidation process can effectively destroy hazardous organic pollutants in air and water streams. Although treatment systems will vary depending on the type of stream being treated, the basic process remains the same. The key ingredient is the photoactive catalyst titanium dioxide (TiO2), which is an inexpensive, non-toxic material commonly used as a paint pigment. When TiO2 is illuminated with lamps or natural sunlight, powerful oxidizing species called hydroxyl radicals form. These radicals then react with the organic pollutant to tear it apart and ultimately form carbon dioxide (CO2) and water (Phillips and Raupp, 1992). When halogenated organics are treated, dilute mineral acids like HCl form. The process works at both ambient and mildly elevated temperatures (>200oC) (Fu et. al., 1995; Falconer and Magrini, 1998). Researchers throughout the world have been investigating PCO as an advanced oxidation technology for treating air and water streams contaminated with a variety of organic and inorganic compounds (Blake, 1996; Cummings et al., 1996). The susceptibility of an organic species to complete oxidation is typically reported in terms of photoefficiency, defined as the number of molecules of contaminant oxidized to carbon dioxide, water, and simple mineral acids divided by the number of photons incident on the catalyst. These values vary widely, depending on the reactor design, catalyst geometry and the compound of interest. Much of the work on photocatalytic oxidation focuses on treating the halogenated organics trichloroethylene (TCE) and perchloroethylene, contaminants commonly found in ground water sources. These compounds and other chlorinated ethylenes typically react rapidly with TiO2 and photons at efficiencies greater than 100%. These rates are likely due to chain reactions propagated by chlorine radicals (Luo and Ollis, 1996; Nimlos et. al. 1993; Yamazaki-Nishida, 1996). Paint solvent emissions generally consist of toluene,

K. A. Magrini, A. S. Watt, L. C. Boyd, E. J. Wolfrum, S. A. Larson, C. Roth, G. C. Glatzmaier 1711

xylenes, ketones and acetate vapors. Measured photoefficiencies for benzene and other aromatic compounds like toluene are typically less than 5% (Gratson et. al., 1995; d'Hennezel and Ollis, 1997). Besides exhibiting low photoefficiencies, aromatic species tend to form less reactive, nonvolatile intermediates during gas phase PCO. These intermediates build up on the catalyst surface and block or inhibit the active catalytic sites for further reaction (Larson and Falconer, 1997). The addition of heat and small amounts of platinum to the TiO2 catalyst overcome these problems (Falconer and Magrini, 1998; Fu et. al., 1996). Oxygenated organics such as ethanol and acetone have photoefficiencies typically around 1-10% (Peral and Ollis, 1992). Several field demonstrations of PCO using sunlight to treat groundwater contaminated with TCE have been reported (Mehos and Turchi, 1993; Goswami et. al., 1993). These field tests found that nontoxic constituents in the water can non-productively react with or “scavenge” the photogenerated hydroxyl radicals and reduce the rate of the desired reaction. Common scavengers such as humic substances and bicarbonate ions increase treatment costs for the technology (Bekbolet and Balcioglu, 1996). Turchi and co-workers, (1994) found that by air stripping the volatile contaminants from the water stream, the regulated compounds at many contaminated sites could be transferred to the air, leaving the radical scavengers behind. The water can then be safely discharged and the air effectively treated with PCO. The improved photoefficiency reduces treatment costs and more than offsets the added cost of air stripping these contaminants from water. Read et. al., (1996) successfully field tested a lamp-driven, PCO system on chloroethylene vapors from a soil vapor extraction well located at DOE's Savannah River Site. Magrini et al., 1996, and Kittrell et al., 1996, both used modified TiO2 catalysts and lamp-driven reactors to treat VOCs representative of semiconductor manufacturing and contact lens production, respectively. 1,2-dichloroethane, stripped from contaminated groundwater, was successfully treated in a pilot scale PCO demonstration at Dover AFB (Rosansky et al., 1998). The second field test assessed PCO to treat paint solvent vapors. Painting operations for military and civilian vehicles are conducted in ventilated enclosures called paint booths. Filters in the exhaust ducts trap paint droplets from the paint overspray while the VOC-laden air is typically exhausted through roof vents. The vent emissions can contain several hundred parts per million (ppm) of the paint solvents, which continue to evaporate from the vehicle after painting is complete. Most types of paint generally contain significant amounts of VOCs such as toluene, a suspected carcinogen, as well as other hazardous solvents such as methyl ethyl ketone, methyl isobutyl ketone, hexanes, xylenes, n-butyl acetate, and other components in lesser amounts. Current technologies for treating these emissions include catalytic combustion of the vapors over supported Cu and Cr-oxides at temperatures of 350oC (Estropov et. al, 1989); air-flow reduction and recirculation strategies (Ayer and Darvin, 1995); and regenerative thermal oxidation at near incineration temperatures (Mueller, 1988). The use of platinized TiO2 and temperatures of 180-200oC provide significant energy savings in treating these emissions. Our goals for testing the paint booth emissions was to gather real-world treatability data and establish that the system maintained performance during the duration of the testing.

1712

21.4 Application of Solar Photocatalytic Oxidation to VOC-Containing Airstreams

Figure 21.4.1. Schematic of the 10 SCFM solar photoreactor system used at McLellan AFB.

21.4.2 PCO PILOT SCALE SYSTEMS 21.4.2.1 Air stripper application Figure 21.4.1 shows a schematic of the pilot-scale system used at McClellan AFB. This system, scaled from an optimized laboratory reactor, was fabricated by Industrial Solar Technology Corporation (Golden, CO). The reactor, 1.22 m wide by 2.44 m, consisted of a fiberglass-reinforced plastic I-beam frame. A transparent fluoropolymer film, treated to accept adhesives, formed the front and back windows of the reactor. The film windows were attached to the reactor frame with foam tape coated on both sides with an acrylic adhesive. The catalyst, titanium dioxide (Degussa P25), was coated onto a structured, perforated polypropylene tubular packing commonly used in oil-water separators. The TiO2 was suspended in water as a slurry and sprayed onto the tubular supports with a new paint sprayer until the supports were opaque. Fluid modeling of airflow through the reactor showed that a 5.1 cm PVC manifold, located at the inlet and outlet of the reactor, would provide an even flow distribution. Small (0.6 cm) holes, drilled into the manifold at one inch intervals, provided the even flow distribution required for efficient contact of the contaminated air with the catalyst. A two-inch diameter pipe from the outlet of the air stripper supplied the contaminated air stream to the reactor. The air stream was first passed through to a tube-in-shell

K. A. Magrini, A. S. Watt, L. C. Boyd, E. J. Wolfrum, S. A. Larson, C. Roth, G. C. Glatzmaier 1713

heat exchanger for partial dehumidification. A chiller circulated 2oC ethylene glycol through the tube side of the heat exchanger. Chilling, which reduced the relative humidity of the air stream from near saturation to less than 20% at 20oC, was required because high humidity reduces the TCE destruction rate, likely because of competitive adsorption between moisture and TCE molecules at the catalyst surface (Fan and Yates, 1996). By lowering the humidity to less than 20% at 20°C, the reaction rate was sufficient to ensure complete TCE destruction. This effect was observed by Hung and Marinas (1997) in a lamp-illuminated annular reactor in which TCE conversion was not affected by relative humidity up to 20%; conversion deteriorated as the humidity level reached saturation. Laboratory tests of the system demonstrated that at TCE concentrations of near 100 ppmv, less than 15% of the TCE in the airstream transferred to the condensate formed in the heat exchanger. The condensate (approx. 1-2 liters per day) could be fed back into the air stripper for treatment. On the reactor, GC sample ports and temperature and pressure sensors provided monitoring of reactor inlet and outlet VOC concentrations and temperature and pressure drop throughout the system. A portable gas chromatograph provided on-line VOC organic analysis of the inlet and outlet air streams. A portable light intensity monitor was used to measure the global horizontal light intensity at the surface of the photoreactor. The calibrated radiometer was mounted on the reactor framework which faced south at an angle of 40o to the horizon. Two blowers in series were used to draw contaminated air into the system. The blowers were located at the end of the air handling system so that the entire system was maintained under slight negative pressure. Because TCE oxidation can produce intermediates such as phosgene, (Nimlos et. al, 1993; Read et. al., 1996; Fan and Yates, 1996), keeping the system under vacuum prevented release of any toxic vapors. The exhaust gases from the blower were returned for final treatment by a caustic contacting tower. Any hydrochloric acid and phosgene formed from TCE oxidation were neutralized and hydrolyzed respectively by the caustic scrubber. The contaminated airstream provided by the air stripper unit contained volatile organic compounds as analyzed from Summa canister tests by Air Toxics, Ltd. using EPA method TO-14. These compounds, listed in Table 21.4.1, consisted primarily of trichloroethylene and trace quantities of dichloromethane, 1,1-dichloroethene, cis-1,2-dichloroethane, chloroform, carbon tetrachloride, and benzene. The other compounds present in the outlet presumably formed during treatment. Table 21.4.1. Results from the Summa canister tests of inlet and outlet reactor air streams during the McClellan tests. The two inlet and two outlet samples were taken sequentially and represent approximately replicate samples. Inlet 1, ppbv

Inlet 2, ppbv

Outlet 1, ppbv

Outlet 2, ppbv

Chloromethane

ND

ND

4.8

4.1

1.5-30

Dichloromethane

36

36

30

32

1.5-30

Compound

Detection limit, ppbv

1,1-Dichloroethane

ND

ND

17

17

1.5-30

1,1-Dichloroethene

200

200

ND

ND

1.5-30

1714

21.4 Application of Solar Photocatalytic Oxidation to VOC-Containing Airstreams

Table 21.4.1. Results from the Summa canister tests of inlet and outlet reactor air streams during the McClellan tests. The two inlet and two outlet samples were taken sequentially and represent approximately replicate samples. Compound

Inlet 1, ppbv

Inlet 2, ppbv

Outlet 1, ppbv

Outlet 2, ppbv

Detection limit, ppbv

cis-1,2-Dichloroethane

220

240

6

2.2

1.5-30

Chloroform

250

260

280

280

1.5-30

Carbon tetrachloride

700

730

790

790

1.5-30

Benzene

170

190

120

110

1.5-30

1,2-Dichloroethane

ND

ND

12

ND

1.5-30

Trichloroethylene

13,000

13,000

660

280

1.5-30

Acetone

ND

ND

84

69

9.8-120

1,4-Dioxane

ND

ND

10

ND

9.8-120

21.4.2.2 Paint booth application For the paint solvent application, a slipstream of a paint booth vent located at Fort Carson was analyzed to determine the components present with rapid scan gas chromatograph. On any operational day, several paint types are sprayed. The solvent emissions content thus varies during painting operations. In general, the amounts and types of solvent emissions from the Fort Carson paint booth are listed in Table 21.4.2. Table 21.4.2. Breakdown of solvent emissions from painting operations at Fort Carson in 1994. VOC refers to volatile organic compounds, HAP to hazardous air pollutants, MEK to methyl ethyl ketone, and MIBK to methyl isobutyl ketone Paint type Solvent base Thinner

Total usage lbs/yr

VOC lbs/yr

HAP lbs/yr

MEK lbs/yr

Toluene lbs/yr

MIBK lbs/yr

Xylene lbs/yr

19,656

110,074

5,405

3,931

983

7,235

722

552

491

0

983

0

174

0

A small field test system, consisting of a photoreactor packed with Degussa TiO2coated glass beads, was used to assess treatability of the slipstream and provide design data for the pilot scale system. The small reactor operated with a flow rate of 20 liters/minute provided by a pump connected to the paint booth exhaust vent line and variable reactor temperature (ambient to 200oC). Inlet and outlet VOC concentrations were simultaneously measured by two gas chromatographs directly connected to both lines. This arrangement provided real time monitoring of the inlet and outlet VOC concentrations during painting operations. The resultant treatability data was used to size a 5-SCFM pilot-scale system (Figure 21.4.2) also built by Industrial Solar Technologies. This system consists of a parabolic trough reflector to focus incident sunlight onto a receiver tube containing the catalyst. The trough has a 91 cm aperture width and is 243 cm long. The reflective surface is covered

K. A. Magrini, A. S. Watt, L. C. Boyd, E. J. Wolfrum, S. A. Larson, C. Roth, G. C. Glatzmaier 1715

Figure 21.4.2. Schematic of the 10 SCFM parabolic trough, solar photoreactor system used at Fort Carson U. S. Army Installation.

with SA-85 (3M Company), a polymer film that reflects greater than 90% of the incident radiation onto the catalyst tube, providing both heat and light for reaction. The catalystcontaining receiver tube is a 3.81 cm I.D. borosilicate tube packed with 1.9 cm ceramic berl saddles coated with 2.0 wt% platinum on TiO2. A higher weight loading of platinum was used in the field test because of the variable VOC composition of the paint booth stream. Operating temperatures of 200oC were obtained, depending on cloud cover, during most of the day. Note that a thermally heated, lamp-illuminated system will provide similar performance as the solar-based system described here. The reflective trough rotates around the fixed receiver tubes by means of a gear motor. A control and photodiode arrangement activate the gear motor, permitting the trough to track the sun throughout the day. Paint booth emissions were sampled directly from the exhaust stream via 10.1 cm aluminum ducting. On-line analysis of the inlet and outlet concentrations was provided by sample lines connected to the MTI gas samplers. Reactor temperature and inlet flows were measured continuously. 21.4.3 FIELD TEST RESULTS 21.4.3.1 Air stripper application The field test at McClellan AFB took place during a 3-week period in April, 1997. System operating conditions included two flow rates (10 and 20 SCFM) and variable light intensity. The photoreactor system, designed to perform optimally at 10 SCFM with natural sunlight, was also operated at 20 SCFM to challenge the system. Because of varying solar illumination levels during the day, the system can operate for longer periods at lower flow rates. Residence times in the reactor at 10 and 20 SCFM were 0.50 sec and 0.25 sec, respectively. Figure 21.4.3 shows TCE conversion and UV intensity versus illumination time at 10 SCFM for two days of operation. UV intensities were measured with a UV radiometer (UHP, San Gabriel, CA, Model UHX). TCE inlet concentrations varied from 10-15 ppmv. Detection limits for the gas chromatograph were measured at 0.5 ppmv. Figure 21.4.3 indicates that conversions greater than 95% at 10 SCFM are achieved with UV intensities at or greater than 1.5 mW/cm2. At 10 SCFM, continuous destruction greater than 95% was achieved for at least 6 hours per day on two clear days in April at this latitude (37°N). The

1716

21.4 Application of Solar Photocatalytic Oxidation to VOC-Containing Airstreams

Figure 21.4.3. Solar insolation (left) and fractional TCE conversion (right) versus illumination time. Flow rate and relative humidity of the incoming air stream was 10 SCFM at 20% RH. Data taken on April 8-9, 1997.

Figure 21.4.4. Solar insolation (left) and fractional TCE conversion (right) versus illumination time during periods of partial cloud cover. Flow rate and relative humidity of the incoming stream was 10 SCFM and 20% RH. Data taken on April 7 and April 16, 1997.

length of operation for a solar-driven system will vary with time of year, latitude, and altitude due to the effect each parameter has on available solar-UV intensity. Figure 21.4.4 shows the effect of clouds on solar insolation (left) and TCE conversion (right) as a function of illumination time at 10 SCFM for two separate days. Again TCE conversions greater than 95% were obtained when the insolation levels were greater than 1.5 mW/cm2. In general, clear skies had UV intensities of 2 mW/cm2 or better. Haze and high clouds reduced the intensity to 1.5 mW/cm2, and thicker clouds reduced available intensity to about 1 mW/cm2. Conversions greater than 90% were still achievable during periods of significant cloud cover due to the photoreactor's ability to use global horizontal (diffuse) UV illumination. TCE conversion versus solar insolation for three days of operation at 20 SCFM is shown in Figure 21.4.5. Doubling the flow rate halves the residence time of the gas in the photoreactor. Conversions greater than 95% are achieved with UV intensities greater than 2.0 mW/cm2. The reproducibility of TCE conversion indicates that the little or no catalyst deactivation occurred during the three weeks of testing.

K. A. Magrini, A. S. Watt, L. C. Boyd, E. J. Wolfrum, S. A. Larson, C. Roth, G. C. Glatzmaier 1717

Figure 21.4.5. Solar insolation (left) and fractional TCE conversion (right) versus illumination time. Flow rate and relative humidity of the incoming air stream was 20 SCFM at 20% RH. Data taken on April 10, 14, and 15, 1997.

Pooling all of the data taken at the two flow rates provides a relation between TCE conversion and UV intensity (Figure 21.4.6). Significant scatter, which exists for both flow rates, is likely due to the rapidly varying UV intensity levels during cloud events. The variable illumination makes it difficult to establish a precise correlation between an isolated UV measurement and the UV exposure received by the gas as it flows through and reacts with the photoFigure 21.4.6. Pooled UV intensity versus TCE conversion data for catalyst bed. The 10 SCFM data all 10 SCFM and 20 SCFM runs. suggests a square root relationship between UV intensity and conversion. The square root dependence was observed in a continuous flow reactor when TCE concentrations were less than 60 ppmv (Nimlos et. al., 1993). Higher concentrations yielded an approximately linear dependence. On the last day of testing, grab samples were collected in Summa canisters and delivered to a local analytical laboratory (Air Toxics, Ltd., Folsom, CA) for TO-14 analysis to confirm the analyses provided by the MTI gas chromatographs and to provide outlet TCE concentrations that were below our GCs detection limits. Replicate samples were reported as 13,000 ppbv for TCE in inlet air, and 280 and 660 ppbv TCE in the outlet (Table 21.4.1). This corresponds to 97.8% and 94.9% removal of TCE. The reactor outlet airstream contained most of the dichloromethane and carbon tetrachloride present in the inlet stream. This result was expected, as single carbon haloorganics do not easily photooxidize (Jacoby et. al, 1994). Nearly all of the 1,1-dichloroethene and cis-1,2-dichloroethane were destroyed (destruction efficiencies of 97.3% and 99.1%

1718

21.4 Application of Solar Photocatalytic Oxidation to VOC-Containing Airstreams

respectively). This too was expected, as dichlorinated ethanes and ethenes photooxidize almost as rapidly as TCE. Small amounts (<20 ppbv) of chloromethane, 1,1-dichloroethane, 1,2-dichloroethane and acetone, and 1,4-dioxane appeared in the reactor outlet and are likely reaction intermediates from TCE oxidation. About 40% of the benzene in the inlet stream did not come out of the reactor. This result is in line with recent results in which benzene was oxidized with a sol-gel TiO2 catalyst (Fu, X. et. al., 1995). Though not all of the benzene reacted, that which did formed only CO2 and H2O. Adding platinum to the TiO2 sol significantly improved benzene oxidation. Transient reaction studies of benzene and Degussa TiO2 thin films indicate that benzene oxidizes rapidly at 300K to form strongly bound surface intermediates that oxidize very slowly (Larson and Falconer, 1997). Other work with thin films of TiO2 and benzene shows that benzene does not form intermediates when water vapor is present. In the absence of water, acetone forms and the catalyst becomes coated with a durable brown material (Sitkiewitz, S. and Heller, A., 1996). Because these tests were run at an average of 20% RH and the catalyst exhibited stable activity, it is likely that benzene was oxidized. Thus, since the catalyst activity did not deteriorate over the course of testing, it may be possible that low concentrations of aromatic species can be tolerated and treated in a solar photocatalytic reactor. It is also possible that the benzene was absorbed by the catalyst support material. 21.4.3.2 Paint booth application Figure 21.4.7 shows the real time test results using the lamp-driven, small field scale reactor for treating VOC emissions from painting operations at Fort Carson. The inlet and outlet VOC concentrations are given in ppm. Test data begins at 8:10 AM on June 25, 1996 and ended at 11:55 for this run. The painting operations are conducted in a batch mode. Vehicles are placed in the paint booth, painted, then removed from the booth to the parking Figure 21.4.7. Inlet and outlet VOC concentrations from area, where they dry. The mixture of VOCs the Fort Carson paint booth using the small, lampemitted from the aliphatic polyurethane driven, field test reactor. green military paint being sprayed consisted of methyl ethyl ketone, methyl isobutyl ketone, toluene, n-butyl acetate, xylenes, and ethyl benzene. When the reactor reached an operating temperature of 170oC, complete VOC destruction was achieved for the entire run except for a period of peak VOC emissions beginning at 9:30 and ending at 9:45. Thus, during this four hour period, the PCO process destroyed virtually all of the VOCs in the slip stream from the paint booth exhaust. The total time averaged conversion is estimated to be greater than 95%. These data were used to design the solar-driven, photothermal reactor system.

K. A. Magrini, A. S. Watt, L. C. Boyd, E. J. Wolfrum, S. A. Larson, C. Roth, G. C. Glatzmaier 1719

Based on the lamp tests, the 5-SCFM thermal photoreactor system was tested with methyl ethyl ketone (MEK), a paint solvent surrogate. Figure 21.4.8 shows the results of on-sun testing with this system. The Y-axis plots three factors: temperature in o Cx100 MEK, conversion as a fraction with 1 equaling 100% conversion, and solar insolation as mW/cm2. The X-axis plots illumination time in minutes. This test Figure 21.4.8. The effect of sunlight fluctuations on the photother- was performed on a hazy day with mal destruction of methyl ethyl ketone in the parabolic trough, pho- numerous clouds to evaluate the tothermal reactor. MEK conversion, insolation level, and reactor system under challenging conditemperature are plotted as a function of continuous outdoor illumitions. Though conversion and nation time. reactor temperature parallel the insolation, they do so in a less than 1:1 ratio. This is likely due to the square root dependence of reaction rate on light intensity and to the heat transfer (insulative) characteristics of the ceramic support. 100% conversion was achieved when insolation levels were above 3.5 mW/cm2 and reactor temperatures above 175oC. The photothermal reactor was tested in July, 1997 for three weeks on paint booth emissions using natural sunlight. A represenFigure 21.4.9. VOC concentration from paint solvents as a function tative data set for July 22, 1997 is of illumination time during photocatalytic treatment with a concen- shown in Figure 21.4.9 in which trating, parabolic trough, solar photoreactor. the total VOC concentration for the inlet and the outlet of the parabolic trough, photothermal reactor is plotted against illumination time. The solvent stream, which varied with time, generally contained about 140 ppmv of toluene, methyl ethyl ketone, methyl isobutyl ketone, xylenes, butyl acetate and ethyl benzene. The operating temperature was from 190 to 208oC, dependent on available illumination. On-sun refers to turning the concentrating receiver into the sun, which initiates heating and photoreaction. Off-sun refers to turning the receiver away from the sun (photoreaction and heating stops). Inlet and outlet VOC concentrations, which matched closely when the reactor was off-sun, were taken to ensure that thermal reaction as the

1720

21.4 Application of Solar Photocatalytic Oxidation to VOC-Containing Airstreams

reactor cooled and adsorption of contaminants onto the catalyst surface were not significant. No attempt was made to decouple photoreaction from thermal reaction because of rapid temperature changes when the trough was rotated off-sun and the difficulty of thermally heating the catalyst. The results of 3 weeks of testing demonstrated that the photothermal treatment system has destruction and removal efficiencies (DREs) in excess of 95% under conditions where incident UV irradiation provided operating temperatures from 150 to 200oC. No loss of performance of the catalyst occurred. Significant results from these pilot-scale field tests: • A non-tracking, ambient-temperature, solar-powered photocatalytic reactor consistently achieved greater than 95% destruction of 10-15 ppmv TCE for 4 to 6 hours of operation per day during April at a latitude of 37°N. Catalyst performance remained constant during three weeks of testing. • 95% destruction was achieved at 10 SCFM with UV intensities greater than 1.5 mW/cm2 and at 20 SCFM at UV intensities greater than 2.0 mW/cm2. • Greater than 95% destruction of a variable paint solvent stream (140 ppm total VOC) was achieved with a concentrating solar photoreactor operated with ambient sunlight at temperatures of 180-200oC. These operating temperatures were achievable for 4-8 hours of operation per day in July at a latitude of 39.5oN. Catalyst performance remained constant during three weeks of testing. 21.4.4 COMPARISON WITH OTHER TREATMENT SYSTEMS The design and fabrication of the photocatalytic reactor system used in the McLellan field test was provided by NREL engineers and the Industrial Solar Technology Corporation (IST), Golden, CO. IST estimates that the cost for a 200-SCFM system is about $14,000.00, depending on the site requirements for hours of operation per day and the site solar potential. Costs to construct the reactors could be significantly reduced through economies of scale and large scale production likely could reduce reactor costs by half. Other purchased equipment costs for the photoreactor include a condenser/chiller and an instrumented blower/scrubber skid system for air handling. The level of instrumentation on the skid directly affects cost. The system described here was instrumented more than would be required for a commercial system. Equipment and installation costs for comparably sized carbon adsorption and thermal oxidation systems were calculated using EPAs HAPPRO v.2.1 software. Thus, total direct costs for the three systems are very close and range from an estimate of $131K for thermal oxidation, $119K for PCO, to $97K for carbon adsorption. Indirect costs were estimated using a factor of 0.28 x total purchased equipment cost. This value includes engineering, construction fee, and start-up costs. (Vatavuk, 1994) Total capital required for all three systems is approximately equal and ranges from $121K to $164K. Operating costs are also similar and range from $23 to $28K and total annual costs range from $37K for thermal oxidation to nearly equivalent values for carbon adsorption and PCO at $32K and $33K, respectively. No fuel is required for PCO and carbon adsorption. Details of the calculations are provided in the footnotes following Table 21.4.3.

K. A. Magrini, A. S. Watt, L. C. Boyd, E. J. Wolfrum, S. A. Larson, C. Roth, G. C. Glatzmaier 1721

Table 21.4.3. Comparison of estimated costs for similarly sized PCO, carbon adsorption, and thermal oxidation VOC treatment systems. Direct Costs

Solar PCO non-concentrating

Carbon Adsorption

Thermal Oxidation

Purchased Equipment (PE) Photoreactor1 Condenser/Chiller2 Instrumented Blower/Scrubber Skid3

14,000 10,625 60,000

Total PE (TPE)4 Installation

91,395 27,419

74,5005 22,350

100,5005 30,150

Total Direct Costs

118,814

96,850

130,650

Indirect Costs Indirect Installation Costs7

29,923

24,391

32,904

Total Capital Required8

148,736

121,241

163,554

Operating Costs9 Electricity Fuel10 Maintenance, Labor and Material

2,880 0 20,376

1,440 0 19,308

1,440 5,000 20,952

Direct Annual Costs Indirect Annual Costs

23,256 8,664

25,748 7,063

27,392 9,527

Total Annual Costs

31,920

32,811

36,919

56,135

25,549

63,456

281

263

318

12

Levelized Cost

Levelized Cost/CFM

1 IST Corp., Golden, CO. Non-concentrating reactor cost is current and estimated price reductions anticipated from 2 Peters and Timmerhaus, 1991. 3 Zentox Corp., Ocala, FL. Concentrating vs. non-concentrating is primarily a function of instrumentation level. 4 Includes 0.08 x PE for taxes and freight (Vatavuk, 1994). 5 Calculated using EPAs HAPPRO v.2.1 software. 6 0.38 x PE (Vatavuk, 1994). 7 0.28 x TPE, including engineering costs, construction fee, and start-up costs (Vatavuk, 1994) 8 Includes 6% contingency. 9 Based on $0.08/kWh, 10 kW blower and chiller costs. 10 Based on $0.33/ccf natural gas. 11 0.06 x Total Capital Required for taxes, insurance, and administration. 12 Based on 10% interest rate, 10 year depreciation.

REFERENCES Ayer, J. and Darvin, C. H., 1995, “Cost Effective VOC Emission Control Strategies for Military, Aerospace, and Industrial Paint Spray Booth Operations: Combining Improved Ventilation Systems with Innovative, Low-Cost Emission Control Technologies”, Proceedings of the 88th Annual Meeting of the Air and Waste Management Association, Vol. 4B, 95-WA77A.02.

1722

21.4 Application of Solar Photocatalytic Oxidation to VOC-Containing Airstreams

Blake, D.M., 1996, Bibliography of Work on the Heterogeneous Photocatalytic Removal of Hazardous Compounds from Water and Air, Update Number 2 to October 1996. NREL/TP-430-22197. Golden, CO: National Renewable Energy Laboratory. Bekbolet, M. and Balcioglu, I., 1996, Water Sci. Techn., 34, No. 9, pp. 73-80. Cummings, M. and Booth, S. R., Proc. Int. Conf. Incineration Therm.Treat. Technol. (1996), pp. 701-713. d'Hennezel, O. and Ollis D. F., 1997, J. Catalysis, 167, No. 1, pp. 118-126. Evstropov, A. A., Belyankin, I. A., Krainov, N. V., Mikheeva, T. Ya., Kvasov, A. A., 1989, “Catalytic Combustion of Gas emissions from A Spray-Painting Chamber”, Lakokras. Mater. Ikh. Primen, Issue 1, pp. 115-118. Falconer, J. L. F. and Magrini-Bair, K. A., 1998, “Photocatalytic and Thermal Catalytic Oxidation of Acetaldehyde on Pt/TiO2”, accepted for publication in the J. Catalysis. Fan, J. and Yates, J. T., J. Am. Chem. Soc., 118, pp. 4686-4692. Fu, X., Clark, L. A., Zeltner, W. A., Anderson, M. A., 1996, “Effects of Reaction Temperature and Water Vapor Content on the Heterogeneous Photocatalytic Oxidation of Ethylene”, J. Photochem. Photobiol. A, 97, No. 3, pp. 181-186. Fu, X., Zeltner, W. A., and Anderson, M. A., 1995, Applied Catalysis B: Environmental, Vol. 6, pp. 209-224. Gratson, D.A.; Nimlos, M.R.; Wolfrum, E.J., 1995, Proceedings of the Air and Waste Management Association. Hung, C.-H. and Marinas, B. J., 1997, Environ. Sci. Techn., 31, pp. 1440-1445. Jacoby, W.A.; Nimlos, M.R.; Blake, D.M.; Noble, R.D.; Koval, C.A., 1994, Environ. Sci. Techn., 28, No. (9), pp. 1661-1668. Kittrell, J. R., Quinlan, C. W., Shepanzyk, J. W., “Photocatalytic Destruction of Hexane Eliminates Emissions in Contact Lens Manufacture”, Emerging Solutions in VOC Air Toxics Control, Proc. Spec. Conf. (1996), pp. 389398. Larson, S. A. and Falconer J. L., 1997, Catalysis Lett., 44, No. 9, pp. 57-65. Luo, Y. and Ollis, D. F., 1996, J. Catalysis, 163, No. 1, pp. 1-11. Magrini, K. A., Watt, A. S., Boyd, L. C., Wolfrum, E. J., Larson, S. A., Roth, C., and Glatzmaier, G. C., “Application of Solar Photocatalytic Oxidation to VOC-Containing Airstreams”, ASME Proceedings of the Renewable and Advanced Energy Systems for the 21st Century, pp. 81-90, Lahaina, Maui, April 1999. Magrini, K. A., Rabago, R., Larson, S. A., and Turchi, C. S., “Control of Gaseous Solvent Emissions using Photocatalytic Oxidation”, Proc. Annu. Meet. Air Waste Manage. Assoc., 89th, (1996) pp. 1-8. Mueller, J. H., (1988), “Spray Paint Booth VOC Control with a Regenerative Thermal Oxidizer“, ProceedingsAPCA Annual Meeting 81st(5), Paper 88/84.3, 16 pp. Nimlos, M. R., Jacoby, W. A., Blake, D. M. and Milne, T. A., 1993, Environ. Sci. Techn., 27, p. 732. Peral, J. and Ollis, D.F., 1992, J. Catalysis, 136, pp. 554-565. Phillips, L.A.; Raupp, G.B., 1992, J. Molec. Catalysis, 7, pp. 297-311. Read, H. W., Fu, X., Clark, L. A., Anderson, M. A., Jarosch, T., 1996, “Field Trials of a TiO2 Pellet-Based Photocatalytic Reactor for Off-Gas Treatment at a Soil Vapor Extraction Well”, J. Soil Contamination, 5, No. 2, pp. 187-202. Rosansky, S. H., Gavaskar, A. R., Kim, B. C., Drescher, E., Cummings, C. A., and Ong, S. K., “Innovative Air Stripping and Catalytic Oxidation Techniques for Treatment of Chlorinated Solvent Plumes”, Therm. Technol., Int. Conf., Rem. Chlorinated Recalcitrant Compounds., 1st (1998) pp. 415-424. Sitkiewitz, S. and Heller, A., 1996, New J. Chemistry, 20, No. 2, pp. 233-241. Turchi, C.S., Wolfrum, E., Miller, R.A., 1994, Gas-Phase Photocatalytic Oxidation: Cost Comparison with Other Air Pollution Control Technologies, NREL/TP-471-7014. Golden, CO: National Renewable Energy Laboratory. Watt, A. S., Magrini, K. A., Carlson-Boyd, L. C., Wolfrum, E. J., Larson, S. A., Roth, C., and Glatzmaier, G. C., “A Pilot-Scale Demonstration of an Innovative Treatment for Vapor Emissions”, J. Air Waste Management, November, 1999. Yamazaki-Nishida, S., Fu, X., Anderson, M. A. and Hori, K., 1996, J. Photochem. Photobiol. A: Chem., 97, pp. 175-179.

ADDENDUM Overview Since the 2nd edition of the Handbook of Solvents was published in 2014, the use of semiconductor photocatalysis via photocatalytic oxidation (PCO) to remove chlorinated organics from air and water streams has further developed as an emerging industry consisting of indoor air cleaning systems, water treatment systems, waste treatment systems, more efficient visible light activated catalysts, and more efficient and varied materials that include

K. A. Magrini, A. S. Watt, L. C. Boyd, E. J. Wolfrum, S. A. Larson, C. Roth, G. C. Glatzmaier 1723

self-cleaning including textiles, ceramics, and mirrors (Ragesh et al., 2014). The range of compounds that are treatable by PCO in water have further expanded from chlorinated organics to include taste and odor compounds, pharmaceuticals including endocrine disruptors (bisphenol A), pesticides, dyes, paints, CECs (contaminants of emerging concern) and microorganisms including pathogenic bacteria (E-coli), fungi, algae and viruses. Photocatalysts continue to evolve from the original photocatalyst, titanium dioxide (TiO2), to novel materials comprising nanoparticles of TiO2, varied metal addition to TiO2 and ZnO, varied types of carbon-doped TiO2, 2-dimensional photocatalysts based on metal oxides, metal chalcogenides and metal free materials such as graphene, and MOFs (metal oxide frameworks) with an emphasis on increasing both ultraviolet and visible light utilization. Nitrogen-doping confers strong visible light responses in TiO2, ZnO and perovskite photocatalysts. Other visible light activated catalysts comprise Nb2O5, plasmonic catalysts based on Au, Ag and Cu supported on metal oxides, and polymeric graphitic carbon nitrides. Combining heterogeneous photocatalysis with molecular photocatalysts offers a new approach to broadening visible-light activated organic transformation. Novel reactor configurations include combinations of photocatalysts with membrane reactors for water treatment and the combination of selective removal of either contaminants or catalyst foulant prior to photocatalytic treatment. Several comprehensive reviews of photocatalysis have been published recently and interestingly conclude that TiO2-based photocatalysts remain the material of choice for varied applications due to its low cost, high oxidizing power and broad applicability to varied contaminants. The following sections describe the significant advances that have occurred in the last five years. Chlorinated organic destruction in air and water streams Photocatalytic water treatment remains at the forefront of PCO research because of the ability of titania and other heterogenous photocatalysts to mineralize/oxidize a wide range of compounds efficiently using low cost reactor designs and facile integration with other advances oxidation processes (AOPs). Several recent reviews catalog progress in destroying chlorinated organics in air and water streams beginning with an expanded compendium of priority water pollutants defined in Directive 2013/39/EU in which AOPs combined with photocatalytic treatment provide an efficient process to destroy persistent pesticides and chlorinated organics (Ribeiro et al., 2015). An emerging aqueous pollutant of concern widely found in the ambient environment is bisphenol A (BPA), which is broadly used as a precursor for epoxy and polycarbonate resins used to coat water containers and medical devices. It is found in water, air and soil and is classified as an endocrine disruptor. Reddy et al. (2018) surveyed the literature on PCO of BPA and concluded that titania performs the best with no production of aqueous toxic intermediates. BPA contaminated water is expected to increase in the coming decades. The dominance of TiO2 for water treatment is interpreted as a focus on bandgap values by the research community and a lack of implementation of modeling tools to design more active photocatalysts (Shaham-Waldman, 2016). Other work reviews modifying ZnO photocatalysts to absorb in the visible spectrum with a focus on destroying persistent dyes and phenols in natural water resources from textile manufacture (Lee et al., 2016). Metal addition, anionic, cationic,

1724

21.4 Application of Solar Photocatalytic Oxidation to VOC-Containing Airstreams

and rare earth dopants have been used to reduce recombination and photo-corrosion and improve efficiency by using ZnO nanoparticles. Taste and odor compounds in drinking water supplies are a current problem that can be addressed by PCO though toxicity of intermediates is not well known and interference by natural organic matter can impact catalyst activity. Non-toxic geosmin and 2-methylisoborneol are drinking water contaminants derived from cyanobacteria and actinomyces that are prevalent and at low concentrations (ng/L) that make treatment difficult. AOPs that treat these species are detailed in a recent review by Antonopoulou et al., 2014. Concern about the presence of biohazards in water supplies such as harmful microorganisms that can lead to epidemic outbreaks and the efficiency of PCO to disinfect these organisms is described in a recent review by Reddy et. al., 2017. TiO2-produced hydroxyl radicals attach to bacterial cell walls and induce cellular damage by rapid leakage of potassium ions. Additionally, peroxide radicals and free metal ions that also form can be effective by damaging the phospholipid layer of bacterial cell walls. Modifications to TiO2 via metal addition and particle size manipulation have treated E. coli, S. aureus, P. aeroginosa and L. monocytogens with treatment times ranging from 1-8 hours. Other aquatic species treatable by TiO2 PCO include the fungi Candia albicans, a virus Bacteriophage MS2 with Ag/TiO2, the protozoa Cryptosporidium and Giardia lamblia, and the algae Chlorella with treatment times of hours and efficiencies of > 80%. Improving catalyst efficiency, designing and deploying larger scale systems, identifying catalyst regeneration and cycle times, and understanding destruction mechanisms still require development if these systems are to be used reliably at larger scales. Nanoparticulate zinc oxide (ZnO) as well has been modified and evaluated for its bactericidal photoactivity. Modifications similar to those applied to TiO2 (metal and non-metal doping, noble metal deposition, and coupling with semiconductors and carbon heterojunction addition) have been evaluated for PCO of aqueous organics and bacteria (S. aureus and E. coli) with nanoparticulate (NP) ZnO exhibiting excellent bactericidal activity via reactive oxygen species generation, Zn+2 release, membrane disfunction and NP internalization though these mechanisms are not completely understood at the cellular level (Qi et al., 2017). New photocatalysts Modifying titania remains a primary research focus to improve its intrinsic photoactivity by doping to extend the photoactive zone. Nanoparticulate titania is becoming more used specifically since it can be immobilized on support systems (Mahlambi et al., 2015) for use in water treatment. Other modification to titania to improve its performance include manipulating catalyst morphology to tune the bandgap to the visible spectrum while reducing recombination (Xu et al., 2014). Anion doping is more effective than cation doping, which generates destructive electron-hole recombination centers (Wang et al., 2018). Nitrogen doping has been successfully used to achieve a high visible light response. Parameters that impact visible light activity include the band gap, available nitrogen doping sites and amounts, defect amounts, and TiO2 particle morphology. Controlling these variables through preparation provides control over the visible light response. Other photocatalysts that benefit from nitrogen doping include ZnO, Ga2O3, and GeO2. Perovskites respond to nitrogen doping as well by forming photoactive perovskite oxynitrides. Reduc-

K. A. Magrini, A. S. Watt, L. C. Boyd, E. J. Wolfrum, S. A. Larson, C. Roth, G. C. Glatzmaier 1725

ing the bandgap of TiO2 and ZnO by metal addition is a straightforward method to increasing visible light activation. Adding nanoparticles of Ag, Fr, Mo and Cu reduces the effective bandgaps of both materials with Ag strongly enhancing photoactivity (Khaki et al., 2017). Doping carbon into the titania structure produces a visible light activated catalyst with carbon replacing oxygen sites and forming oxygen vacancies. At >400 nm illumination, 23% NO destruction was achieved compared with 60% at 290 nm illumination (Wu et al., 2013). Graphite doping into titania achieved a similar result with visible light activation at 420 nm greater than pure anatase TiO2 for methyl orange destruction, 35.8 vs. 21.3%. respectively (Jia et al., 2016). Other work used low cost starch as a carbon source to dope pure anatase to produce a 2.79 eV, which is within the visible portion of the spectrum. Rhodamine B dye destruction with the doped material was more efficient than either undoped titania or Degussa P-25 titania (Warkhade et al., 2017). Another novel carbon doping approach uses an in-situ synthetic strategy to produce carbon-TiO2 nanotubes by calcining surface sulfonated polystyrene nanofibers/TiO2 composites. The resulting 200 nm diameter tubes were more active under visible and UV light than the base titania with the tube morphology more applicable to waste water treatment applications (Ji et al., 2017). Zinc oxide nanostructure development as an inexpensive alternative to titania is a current research focus with dopant addition (nonmetals comprising N, C, and S; metals including transition, rare-earth and aluminum which substitutes for Zn and red shifts the band gap) used to improve visible light activity (Samadi et al., 2016; Lee at al., 2016). Interestingly, most of these studies evaluated PCO activity with model compounds; not much work has been done with actual waste streams. Another type of new photocatalysts are two dimensional (2D) materials consisting of three primary types: metal oxides, metal chalcogenides, and metal free photocatalysts tailored for PCO. Photoactive metal oxide examples include TiO2, WO3, titanoniobate, and perovskite oxides; chalcogenide nanosheets of MoS2, SnS2, TiS2, WS2, MoSe2; and metalfree materials based on light-weight elements comprising carbon, phosphorous, binary carbon nitride, hexagonal boron nitride and boron carbide. Graphene additionally is of interest because of its high carrier mobility (Luo et al., 2016). Further development of these PCO candidates are required before they are useable in actual applications. Silver orthophosphate (Ag3PO4) is a new visible light activated photocatalyst that due to its low bandgap utilizes >90% of wavelengths >420 nm. This catalyst however requires a sacrificial agent as it can photo decompose. Adding Ag nanoparticles overcomes this tendency (Dewil et al., 2017). New applications Pharmaceuticals as emerging pollutants have become a major concern not only because of the threat to health and safety of aquatic species but also due to their continuous accumulation in aqueous environments and the development of antibiotic-resistant microbial strains (Ganiyu 2015). Single water treatment techniques are largely ineffective but combining processes like membrane separation with AOPs including PCO can be completely effective as each process overcomes the challenges of the other. Coupling PCO with membranes is attractive because of simultaneous separation and destruction of pollutants, selfcleaning/anti-fouling properties by foulant destruction, and flexibility for commercial

1726

21.4 Application of Solar Photocatalytic Oxidation to VOC-Containing Airstreams

applications. Generally, PCO combined with membrane technology can either be a pre- or post-treatment stage or a hybrid/one-pot process (Zheng et al., 2017). An example is the complete destruction of the drugs Tamoxifen and Gemfibrozil in a side pressurized and submerged membrane photocatalytic reactor with crystalline TiO2 after 20 minutes with >90% total organic carbon (TOC) removal in 120 minutes. An economic disadvantage to large scale deployment of these hybrid reactors is UV-light source cost. Perspectives on AOPs Dewil et al., 2017 provide a succinct review of advanced oxidation processes including PCO and notes “AOPs should not and cannot do the “full monty” in that they must select specific treatment targets in which they excel: microcontaminant removal, pathogen destruction, and polishing treated industrial streams”. Also required is taking the next step from academic and laboratory scale studies to large-scale, fully operational applications. Byrne et al., 2018 also review recent advances in photocatalysis for environmental applications. Coupling AOPs with renewable energy is critical for simultaneous water treatment and energy production. Additionally, waste valorization coupled with AOPs can reduce process costs; an example is using red mud, a process waste solid from bauxite processing, as an AOP catalyst as it contains useable titania, alumina and iron oxides. A recent review of chlorinated volatile organics (VOCs) destruction by AOPs (Dai et al., 2018) determined that PCO works well at total VOC contents < 500 ppm. Advancements in applying PCO to mitigate VOCs in indoor air environments (Huang et al., 2014, 2016) environments highlight the need to develop visible light activated materials and efficient regeneration mechanisms as products of incomplete oxidation can be more toxic than the parent species and deactivate catalyst surfaces that in turn reduce performance. In summary, significant progress has been made in the development of PCO for treating chlorinated and other organics in water streams since the last update in 2013. Of note is the development of efficient visible light absorbing photocatalysts, two dimensional catalysts, nanomaterials, and coupling with other processes that either provide targeted destruction of specific compounds or separation of recalcitrant species that can be treated with other methods. Bactericidal activity continues to expand with viruses, algae, fungi found to be amenable to photocatalytic oxidation as are persistent odor compounds in drinking water. Numerous recent reviews catalog PCO development with most noting that the time has come to develop applications for actual use in larger-scale systems. REFERENCES Antonopoulou, M., Evgenidou, E., Lambropoulou, D., Konstantinou, I., "A review on advanced oxidation processes for the removal of taste and odor compounds from aqueous media", Water Research 53 (2014) 215-234. Byrne, C., Subramanian, G., Pillai, S. G., "Recent advances in photocatalysis for environmental applications", J. of Environmental Chemical Engineering 6 (2018) 3531-3555. Dai, C., Zhou, Y., Peng, H., Huang, S., Qin, P., Zhang, J., Yang, Y., Luo, L., Zhang, X., "Current progress in remediation of chlorinated volatile organic compounds: A review", J. of Ind. And Eng. Chem. 62 (2018) 106119. Dewil, R., Mantzavinos, D., Poulios, I., Rodrigo, M. A., "New perspectives for advanced oxidation processes", J. of Environmental Management 195 (2017) 93-99. Ganiyu, S. O., van Hullebusch, E. D., Cretin, M., Esposito, G., Oturan, M. A., Coupling of membrane filtration and advanced oxidation processes for removal of pharmaceutical residues: A critical review", Separation and Purification Technology 156 (2015) 891-914.

K. A. Magrini, A. S. Watt, L. C. Boyd, E. J. Wolfrum, S. A. Larson, C. Roth, G. C. Glatzmaier 1727 Huang, B., Lei, C., Wei, C., Zeng, G., "Chlorinated volatile organic compounds (Cl-VOCs) in environment sources, potential human health impacts, and current remediation technologies", Environment International 71 (2014) 118-138. Huang, Y., Ho, S. S. H., Lu, Y., Niu, R., Xu, L., Cao, J., Lee, S., "Removal of indoor volatile organic compounds via photocatalytic oxidation: A short review", Molecules 21 (56) (2016) 1-20. Ji, L., Zhang, Y., Miao, S., Gong, M., Liu, X., "In situ synthesis of carbon doped TiO2 nanotubes with an enhances photocatalytic performance under UV and visible light", Carbon 125 (2017) 544-550. Jia, J., Li, D., Wan, J., Yu, X., "Characterization and mechanism analysis of graphite/C-doped TiO2 composite for enhanced photocatalytic performance", J. of Industrial and Engineering Chemistry 33 (2016) 162-169. Khaki, M., Shafeeyan, M., Raman A., Daud, W., "Application of doped photocatalysts for organic pollutant degradation - A review", J. of Environmental Management 108 (2017) 78-94. Lee, K. M., Lai, C. W., Ngai. K. S., Juan. J. C., "Recent developments of zinc oxide based photocatalyst in water treatment technology: A review", Water Research 88 (2016) 428-448. Luo, B., Liu, G., Wang, L., "Recent advances on 2D materials for photocatalysis", Nanoscale 8 (2016) 69046920. Mahlambi, M. M., Ngila, C. J., Mamba, B. B., "Recent developments in environmental photocatalytic degradation of organic pollutants: The case of titanium dioxide nanoparticles - A review", J. of NanoE., (2015) 1-29. Qi, K., Cheng, B., Yu, J., Ho, W., "Review on the improvement of the photocatalytic and antibacterial activities of ZnO", J. of Alloys and Compounds 727 (2017) 792-820. Ragesh, P., Ganesh, V. A., Nair, S. V., Nair, a. s., "A review on self-cleaning and multifunctional materials", J. Materials Chemistry A2, (2014) 14773-14797. Reddy, P. (a), Kavitha, B., Reddy, P., Kim, K-H., "TiO2-based photocatalytic disinfection of microbes in aqueous media: A review", Environmental Research 154 (2017) 296-303. Reddy, P. (b), Reddy, P, Kwon, E., Kim, K-H., Akter, T., Kalagara, S., "Recent advances in photocatalytic treatment of pollutants in aqueous media", Environment International 91 (2016) 94-103. Reddy, P.(c), Kim, K-H., Kavitha, B., Kumar, V., Raza, N., Kalagara, S., "Photocatalytic degradation of bisphenol A in aqueous media: a review", J. of Environmental Management 213 (2018) 189-205. Ribeiro, A. R., Nunes, O. C., Pereira, M. F. R., Silva, A. M. T., "An overview on the advanced oxidation processes applied for the treatment of water pollutants defined in the recently launched directive 2013/39/EU", Environment International 75 (2015) 33-51. Samadi, M., Zirak, M., Naseri, A., Korashadizade, E., Moshfegh, A., "Recent progress on doped ZnO nanostructures for visible-light photocatalysis", Thin Solid Films 605 (2016) 2-19. Shaham-Waldmann, N. and Paz, Y., "Away from TiO2: A critical minireview on the developing of new photocatalysts for degradation of contaminants in water", Materials Science in Semiconductor Processing 42 (2016) 7280. Warkhade, S., Gaikwad, G. S., Zodape, S. P., Pratap, U., Maldhure, A. V., Wankhade, A. V., "Low temperature synthesis of pure anatase carbon doped titanium dioxide: An efficient visible light active photocatalyst", Materials Science in Semiconductor Processing 63 (2107) 18-24. Wang, W., Tade, M. O., Shao, Z., "Nitrogen-doped simple and complex oxides for photocatalysis: A review", Progress in Materials Science 92 (2018) 33-63. Wu, X., Yin, S., Dong, Q., Guo, C., Li, H., Kimura, T., Sato, T., "Synthesis of high visible light active carbon doped TiO2 photocatalyst by a facile calcination assisted solvothermal method", App. Cat. B: Environmental 142-143 (2013) 450-457. Xu, H., Ouyang, S., Liu, L., Reunchan, P., Umezawa, N., Ye, J., "Recent advances in TiO2-based photocatalysis", J. of Materials Chemistry A 2 (2014) 12642-12661. Zheng, X., Shen, Z-P., Shi, L., Cheng, R., Yuan, D-H., "Photocatalytic membrane reactors (PMRs) in water treatment: Configurations and influencing factors", Catalysts 7 (2017) 224