Control VOCs in Refinery Wastewater

By Mike Worrall and Irl Zuber

Paper presented at the Process Optimization Conference
Houston, TX - March 1998

Oil and water do not mix, like many of life's one-liners this statement is basically true but not the whole story. Many hydrocarbon liquids, particularly aromatics, have significant solubilities in water:

  • Benzene 1800 PPMV
  • Toluene 470
  • Ethyl Benzene 150
  • Xylenes 150

Petroleum refineries do not like salts in their feedstock since these corrode and foul process equipment. The first refining step is desalting where a hot water wash extracts the salts. If feedstock contains aromatics then some will be in the desalter effluent and this is a major source of refinery wastewater containing VOCs.

Usually the desalter is the major source of contaminated process wastewater and typically also has the highest BTEX content. At several refineries the desalter effluent flow has been as high as 50% of the total wastewater flow and over 70% of total BTEX discharge.

Benzene Recovery unit
Closed-loop vapor recovery unit reduces HAP/VOC emissions at Hawaiian refinery

The environmental community is concerned about releases of VOCs and HAPs (Hazardous Air Pollutants) to rivers and streams, to groundwater sources, as well as to the atmosphere. Since aromatics, such as benzene, are considered potential carcinogens, they have received considerable regulatory attention, and are classed as HAPs as well as VOCs. The National Emission Standards for Hazardous Air Pollutants (NESHAPs) require that discharges containing more than ten metric tons per year of a HAP, such as benzene, are subject to regulation - that's an average of only 2.5 lbs/hr, above this threshold stringent levels of control are required. If other HAPs are also present then these also have to be controlled. For benzene discharges regulators require control device efficiencies exceeding 99%.

Other processing units are also sources of aromatics in the process wastewaters. Chemicals units producing aromatics being prime examples. Aromatics are totally soluble in other hydrocarbons and only partially soluble in water. Typical benzene in water levels are 20 PPMW to 200 PPMW, and dependent on feedstock other aromatics may be present in similar amounts.

The main effluent treatment facility often includes an activated sludge unit where bio-degradation converts the final traces of aromatics and other HAPs in the wastewater to carbon dioxide and water.

NESHAPs do not permit open process drains since HAPs could evaporate into the atmosphere prior to reaching the wastewater treatment facility. So it is necessary to provide separate closed drainage systems for HAPs contaminated wastewater. As enclosure of drainage systems is extremely expensive the HAP treatment unit is often located adjacent to the HAPs source.

Just as refineries vary in size so do HAP contaminated process wastewater flows: from 100 GPM at a small refinery to over 3,000 GPM at a large complex. A 500 GPM flow containing 50 PPMW benzene is an annual benzene discharge of 54 tons and therefore subject to regulation.

There are several techniques available to prevent or control HAPs and VOCs in wastewater discharges. These are described and evaluated below:

  • Desalter Emulsion Breaker: The desalter water wash produces an emulsion that holds more benzene and other aromatics than water and if the emulsion is discharged with the wash water it increases the aromatics discharge. Desalters use heat, electrical fields and demulsifiers to minimize the emulsion. Dependent upon feedstock chemistry it can be advantageous to increase demulsifier usage or change demulsifiers to reduce the amount of emulsion discharge. One recent trial reported that changing demulsifier reduced benzene discharge by 50%.
  • Activated Carbon: Direct treatment of the wastewater with activated carbon reduces aromatics content to below acceptable limits. In addition, the carbon also captures oil, grease and other organics. Working capacity of carbon in the liquid phase is about 5% of carbon weight - the spent carbon is returned to the carbon factory for high temperature kiln regeneration and reuse. Although effective, the operating costs are high. One study found that to treat 500 GPM of wastewater entailed $250,000 capital cost and annual operating costs exceeding $1,200,000 (freight to/from kiln, kiln fuel, carbon make-up, etc.).
  • Steam Stripping: Bringing wastewater to the boil by live steam injection effectively strips volatiles such that discharge contains less than 0.5 PPMW aromatics. If overheads condensate comprises equal amounts of aromatics and water they will phase separate: with over 95% of the hydrocarbons in upper phase which recycles to the refinery feedstock. Aqueous phase, with solubility levels of organics, recycles to stripper for cleanup.
    Steam stripping has several concerns: fouling of equipment with oil/grease: fouling of packing with salts, particularly those that precipitate at stripper operating temperatures: energy consumption, even with 75% heat recovery a 500 GPM unit requires 25,000 lbs/hr of steam: capital cost is substantial since stripping column diameter exceeds 10 feet.
  • Air Stripping: Stripping wastewater with air is very effective and readily reduces total BTEX to less than the required 0.5 PPMW. Air stripping is best at around 100 degrees F. As temperature drops packing height increases - at 60 degrees F required packing height doubles to attain same discharge. Typical stripping air discharges contain 500 to 3000 PPMV aromatics and environmental regulations require aromatics capture before air is discharged. The VOC laden stripping air is passed through vapor phase carbon which retains the organics allowing cleansed air discharge. Note that, in the vapor phase, carbon holds several times the quantity of VOC held by liquid phase carbon. There are two carbon options: off-site or on-site regeneration. Off-site regeneration entails shipping the VOC laden carbon to a kiln for high temperature regeneration. On-site regeneration entails live steam desorption of the carbon - usually a bed requires steamout once a shift.
    Areas of concern for air stripping are: safety, since in refinery upset conditions large quantities of hydrocarbons may get into the wastewater resulting in explosive conditions in air stripper and vapor phase treatment unit: fouling of air stripper packing with oil/grease: fouling of packing with compounds that precipitate, particularly those that react with oxygen: fouling of carbon by hydrogen sulfide, note that in an oxygen free situation carbon has very limited capacity for hydrogen sulfide, however, presence of oxygen enables chemisorption onto the carbon as elemental sulfur that fouls the adsorption pores thereby decreasing capacity for aromatics and other VOCs.

An improvement has been developed that utilizes the advantages of air stripping and addresses the concerns listed above. The improvement was conceived and patented by Texaco who worked with the carbon adsorption systems engineers of AMCEC to develop full scale units which AMCEC provides on an exclusive worldwide basis.

Nitrogen is used as stripping gas thus inerting the process. Since oxygen is not present the safety issue is answered. Lack of oxygen (typically now well below 1%) inhibits many concerns about salt and other foulants precipitating on to packing, particularly biological slime formation. Also lack of oxygen reduces chemisorption of hydrogen sulfide onto the carbon thereby extending its working life. Obviously nitrogen is expensive (a 500 GPM wastewater flow requires a stripper gas flow rate of about 2,000 SCFM) therefore, the cleansed gas from the carbon beds is recycled to the stripper.

On-site regeneration was selected. Live steam is used for regeneration in such a manner that carbon vessels (adsorbers) are not isolated from the nitrogen stripping loop with the subsequent need to purge with nitrogen after each steamout to ensure that oxygen is not present. On-site regeneration also avoids frequent transportation of spent carbon to the kiln for reactivation.

Since system is a Recovery process it is considered a process unit and therefore does not require the stringent permitting associated with hazardous waste units. Thus a win-win process has been developed.


Improved process now known as AMCEC BRU (Benzene Recovery Unit)

CASE STUDY

About a dozen BRU systems are operational at various refineries. A typical system is described below:

  • Wastewater Flow 500 GPM
  • Temperature 90 to 130oF
  • BTEX 100 PPMW
  • Hydrogen Sulfide 0 to 5 PPM
  • Required BTEX removal to < 0.5 PPMW

HAP contaminated wastewater from an API separator is pumped into stripping tower and descends through 20 feet of high efficiency packing - 2.5 inches diameter polyethylene open type spheres. The stripped water discharges with less than 0.5 PPMW BTEX so meeting the water discharge regulations.

Nitrogen circulates at 2000 SCFM so that BTEX concentration at top of stripper is about 200 PPMV. Nitrogen temperature equalizes to that of incoming wastewater and exits stripper at 100% relative humidity (RH). For effective carbon adsorption the VOC laden gas (VOCLG) must be at less than 50% RH and as cool as practical.

Adsorption is temperature sensitive and carbon rapidly loses working capacity above 120 degrees F. Moisture content of the VOCLG is important as will be seen from the graph:

With RH levels below 40% carbon has a slight affinity for water. Above 60% RH the capacity to hold water increases tenfold and water is attracted to the same adsorption sites as organic compounds. Because of the vast differences in heats of adsorption between water and organics (1350 compared to 140 BTU/lb), it is not possible to displace water with an organic unless substantial external energy is provided. Thus with high RH conditions carbon may become "water logged" and unable to capture organics. Therefore, VOCLG is cooled to about 90 degrees F which condenses much of the water vapor. After demisting VOCLG is heated to 115 degrees F lowering RH to 40%. Thus VOCLG enters carbon at a suitable temperature and RH for effective adsorption.

Two carbon beds are provided - each bed sized to adsorb for an eight hour shift at full incoming BTEX load. The carbon bed captures more than 99% of the BTEX so the nitrogen recycled to the stripper contains less than 20 PPMV. The stripper had been sized to operate with 50 PPM BTEX in the nitrogen stripping gas so the 20 PPMV does not upset stripping performance. While one carbon adsorber captures BTEX the other is being counter-flow live steam regenerated. Desorption steam displaces the nitrogen in the adsorber being regenerated pushing it through the condenser into the main nitrogen loop - raising loop pressure from 2 PSIG to 7 PSIG which does not impact stripper or adsorber performance and avoids loss of nitrogen. Steam slowly heats the carbon bed to about 230 degrees F releasing much of the adsorbed BTEX. The steam/BTEX vapors flow to a heat exchanger for condensation and cooling. Provided steam flow is within the condenser size and coolant flow there is no passage of steam or BTEX vapor past condenser and into main nitrogen loop.

At completion of desorption steam flow ceases and as steam in desorbing vessel cools, nitrogen is drawn back through the condenser from the main nitrogen loop. This reduces loop pressure back to 2 PSIG. Desorbed vessel is cooled by a slip stream flow of nitrogen from main loop, after 30 minutes bed is sufficiently cool for return to adsorption service.

Regeneration is completed within 4 hours so the adsorber is then parked in standby mode until the on-line adsorber needs to be regenerated. The long idle period permits the BRU system to handle substantial surges in VOC inflow and also accommodate some loss of carbon activity without upsetting overall system performance. An analyzer monitors the nitrogen leaving the carbon bed for VOC content and will initiate early desorption of the adsorber if it is overloaded.

Condensed steam and BTEX are decanted into organic and aqueous phases. Organic layer is pumped to refinery feed. Aqueous phase, containing about 1000 PPMW organics, is pumped into wastewater feed entering the stripper.

Steam consumption is around 1500 lbs/hr. Electric power including that for the air cooled dehumidification and desorption condensing units, but excluding wastewater pumps, is about 50 Kw. Nitrogen consumption is about 5 SCFM which is mainly used by the blower shaft seals.

Capital cost for equipment and controls built to refinery standards, excluding foundations and installation, was about $1,250,000.

Operating experience has been good with little system downtime. Carbon beds (each 5,000 lbs) are replaced every six to twelve months, as pores are fouled by higher boiling compounds and elemental sulfur from the chemisorption of hydrogen sulfide. Refinery wastewater entering the BRU has passed through an API separator and a dissolved air floatation unit so minimizing oil and grease content. Nevertheless on one or two occasions the BRU has received quantities of oil that temporarily coat the stripper packing reducing its performance. Within a few hours the wastewater washes the packing clean restoring full performance.

Hydrogen Sulfide: at facilities where there is a constant hydrogen sulfide load of more than a few PPM it is advisable to add to the BRU treatment train. One such project had about 25 PPM hydrogen sulfide in the wastewater (6 lbs/hr in 500 GPM). Refinery wanted most of the H2S removed and it was preferred that the period between changeout of the regenerable carbon bed be at least six months.

Most of the H2S strips out with the BTEX. The VOCLG then passes to a caustic scrubber where most of the H2S is captured by the scrub liquor. Scrub liquor is a 10% sodium hydroxide solution enroute to another process unit: so the caustic solution only made one pass through the scrubber before discharge to the other process unit. After scrubbing the VOCLG passes through a guard bed of impregnated carbon before entering the regenerable carbon adsorbers. The impregnated carbon chemisorbs most of the remaining H2S - bed is non-regenerable and is replaced every six months.

In summary the BRU system has proven an effective and dependable means to remove BTEX from process wastewaters answering the safety concerns of refinery people, yet compared to the alternatives, has low operating and capital costs.

Authors:

Mike Worrall
A British educated engineer with over 30 years experience in the design of carbon adsorption systems for industrial clients. Mike is Vice President of AMCEC Inc.

Irl Zuber
A Chemical Engineering graduate from Purdue University with 30 years experience of process development and project management for the alternate energy and refining industries. Irl is a senior technologist with Texaco Refining and Marketing at their Bakersfield, California refinery.