Control System for Odorous Emissions from Rock Media Trickling Filters
The South Monmouth Regional Sewerage Authority's 9-mgd wastewater plant had been identified as a source of odors that bothered neighbors from time to time. As a result of work done on the plant's two trickling filters, which were shown to be the culprits generating the unpleasant odors, emissions from the filters have been significantly reduced and complaints of odors have been virtually eliminated.
This New Jersey wastewater authority (SMRSA) owns and operates approximately 15 miles of collector sewers, 11 pump stations and force mains, and the treatment plant. Flow through the plant currently averages about 5.5 mgd, and the facilities serve the sewerage needs of eight municipalities located along the northern part of the New Jersey shore.
Treatment plant process components are: a mechanical bar screen, two aerated grit chambers, two primary settling tanks, a settling tank effluent pump station, two trickling filters, a filter effluent pump station, two final settling tanks, a final settling tank effluent pump station and sludge return pump station (to grit chamber), and an aerated stabilization pond. The sludge is anaerobically digested and hauled for disposal off-site as either liquid, or as cake after dewatering by vacuum filter.
Complaints going back to 1987 from plant neighbors had prompted the New Jersey Department of Environmental Protection (NJDEP) to issue fines to the Authority totalling $172,000 by 1991. On September 20 of that year the Authority and the NJDEP negotiated an Administrative Consent Order (ACO), which lowered the fines to $60,000 and required the Authority to complete specific steps toward controlling odorous emissions from the trickling filters by certain dates.
The Trickling Filters
Each trickling filter (see the magazine cover for a cutaway drawing of a modified unit) is 184 feet in diameter with a 7-ft bed of rock media. Wastewater is distributed uniformly over the rock surface by a rotating arm. An underdrain system of filter blocks supports the rock and collects the filter effluent, which flows toward two parallel channels that run across the filter. These channels have inspection manholes at each end for access and for ventilation. Effluent leaves the unit through the downstream manhole and passes to the filter effluent pump station. A series of 4-in.-diameter vent pipes, spaced every 6 ft around the perimeter of the filter, are also part of the ventilation system.
The filters can be operated in series or in parallel. At present they are running in series, with No. 1 as the primary filter and No. 2 as the secondary unit. In this mode of operation the treated wastewater is collected in the primary filter's underdrain and flows by gravity to the pump station. From there it is pumped back to the intermediate settling tank, then, by gravity, to the secondary filter. Its effluent passes by gravity to a separate wet well in the effluent pump station, then is pumped to the final settling tank.
Trickling Filter Process
A trickling filter is an aerobic biochemical process system in which microorganisms growing on the surface of the rock media consume the biodegradable substances in the wastewater. These microorganisms require a supply of oxygen to perform their treatment function.
Trickling filters rely on natural draft ventilation to provide the oxygen required by the microorganisms. The driving force for natural draft ventilation is the difference in weight between the air in the filter and the air outside. The cooler the air, the higher its density. Depending on the relative temperature of the ambient outside air and that of the wastewater trickling through the rock media, the air movement through the filter will be upward, downward, or stagnant.
During the daytime hours of a typical summer day, the ambient air is warmer than the wastewater, which will cool the air around the filter media, making it more dense than the outside air. This results in the denser air flowing down through the filter media and out of the filter through inspection manholes, underdrain vent pipes, or through the filter effluent pipe above the liquid to the next treatment process.
The opposite is true during a typical winter day, when the ambient air may be cooler than the wastewater trickling down through the filter media. In this situation the wastewater will warm the air around the media and make it less dense than the outside air. This results in the lighter air flowing upward through the media and out of the filter.
There are periods during which there is little or no air flow through the filter in either direction. It has been estimated that when the temperature difference between the ambient air and the wastewater is less than 12 to 16 F, an inadequate air supply through the filter results, and anaerobic conditions within the filter can occur. The air flow may reverse direction in a trickling filter as often as twice per day; in the evening when the ambient air cools and in late morning when the ambient air warms up. In addition to generating odors, periods of inadequate air flow can result in loss of process efficiency. When air movement resumes, either upward or downward, odorous gases may be released to the atmosphere. There is nothing the plant operator can do to control the natural draft ventilation of the filter.
This trickling filter behavior coincided with the pattern of odor complaints from the SMRSA plant's neighbors. The record showed that complaints were mainly registered during periods of stable meteorological conditions and low wind velocities.
It is not unusual to provide forced ventilation to trickling filters that use plastic media. This is because plastic media filters are typically very deep (20 to 40 ft) and highly loaded, and consequently cannot rely on natural draft to provide the necessary oxygen supply. Jacobs Environmental, Inc. (JE) suggested to the Authority that it investigate the feasibility and cost-effectiveness of force-ventilating the trickling filters. There seemed to be no reason why the concept could not be adapted to a trickling filter using rock media, although to the best of our knowledge this had never been done.
In the SMRSA case it offered two major benefits. One is that it would ensure adequate air supply through the rock media at all times. This would prevent anaerobic conditions from developing and enhance the treatment process. Another advantage, from an odor control point of view, is that forced ventilation, in the proper configuration, could develop a negative pressure within the filter, which would permit the capture of odorous gases released there. The gases could then be treated for odor before being exhausted to the atmosphere. Consequently, there would be no need to construct an enclosure over the filter to contain the gases for later deodorization.
The SMRSA plant's average daily flow is about 5 mgd. BOD values run around 250 mg/1. On the basis of a 3 percent oxygen transfer efficiency, on average the trickling filters would require 12,000 cfm of air to provide 85 percent BOD removal. In a forced ventilation system designed to create a negative pressure within the media, JE recommended an air flow rate of about 1 cfm/sq ft of filter area-or about 35,000 cfm per filter. The consultant proposed the construction of an air handling system capable of providing 25,000 cfm and 40,000 cfm of air per unit. This flexibility would allow using the higher rate during the summer and during periods of stable meteorological conditions and low wind speeds, and lower air flow rate during winter, when odor complaints are less likely, and excessive cooling of the wastewater is not desirable.
In late May of 1991, the engineers conducted smoke testing of Trickling Filter No. 1, which is the one more heavily loaded and therefore more likely to plug with biofilm. The test was designed to verify that there was sufficient void space for ventilation air to flow through the filters. This would ensure there was no physical reason that forced ventilation would not work. Several conclusions were reached.
- The filter media and underdrain appeared to be clear, as smoke was visible rising across the filter.
- A heavy volume of smoke was observed rising through the rock around the center column, which indicated short-circuiting. This could be minimized by installing a baffle around the center column.
- There was little evidence of smoke near the filter's perimeter wall, in the areas perpendicular to and farthest away from the central effluent channels. To be certain of achieving adequate ventilation in those areas, the underdrain space at the quarter points could be ventilated by inducing air flow through several of the 4-in.-diam vent pipes.
Pilot Testing Scrubbers
Pilot tests were conducted during July of 1991. They were designed to determine the type of scrubber technology, combination of chemicals, and feed rates that would produce the best odor removal from the trickling filter off-gases. The suction of the pilot unit fan was connected to the downstream manhole of Filter No. 1 with several sections of pipe to transfer the off-gases to the first stage scrubber. Packed tower-type and mist-type scrubber combinations were tested using various dosages of sodium hypochlorite, sodium hydroxide and water.
On July 25, as part of the gas sampling and odor-panel analyses performed for the pilot tests, the gas sample from the primary filter's effluent manhole measured a dilution-to-threshold value (D/T) of 615. By definition, a D/T value of 1 can only be detected by 50 percent of the population and is defined as the odor detection threshold. Therefore, a D/T of 615 would have to be diluted with odor-free air 615 fold for it to reach the detectability threshold concentration.
In addition to the odor panel tests, the 11 samples collected that day were subjected to sulfur compound analysis using gas chromatography/flame photoionization detection (GC/FPD). The results indicated all samples contained hydrogen sulfide, sulfur dioxide and carbon disulfide at parts per billion (ppb) levels. The sample from the filter effluent manhole also was tested using gas chromatography/mass spectroscopy (GC/MS). Hydrogen sulfide was the only compound measured at a concentration above reported odor threshold concentrations.
Based on the pilot test, the following conclusions were reached:
- The fine-mist scrubber performed better than the packed tower.
- A packed tower as a second-stage scrubber to polish the outlet gas from the fine-mist scrubber did not result in further odor reduction .
- A fine-mist scrubber using a sodium hypochlorite and sodium hydroxide scrubbing solution could reduce the odor level at the scrubber stack to below a 100 D/T value.
- The odor level of the outlet gas would have to be further diluted through atmospheric dispersion to achieve the required odor control.
Odor Dispersion Modeling
Modeling was performed to determine the stack height needed to disperse the scrubber exhaust to concentrations below detection levels at the neighboring receptors. The calculations indicated that a 40,000 cfm scrubber exhaust with a 100 D/T odor level would require a 50-foot-high stack. Offsite odors would be below the detection level, even under stable atmospheric conditions and a 1 m/sec wind speed.
Air Pollution Control Permit
Hydrogen sulfide (H2S) had been detected during pilot testing. Also, particulates (salts) result from the reaction of sodium hypochlorite with oxidizable compounds and from unreacted sodium hypochlorite and sodium hydroxide feed to the scrubbers. Therefore, the original application to NJDEP for a Permit to Construct and Certificate to Operate the air pollution control system was based on controlling the emissions of H2S and particulates.
The agency reviewed the permit application and required that in addition to H2S and particulates, the permit include total volatile organic substances (total VOSs), toxic VOSs, dimethyl sulfide (DMS), carbon disulfide (CS2), ammonia (NH4) and amines. To confirm the presence of these additional compounds in the wastewater, one grab sample of trickling filter influent and one grab sample of effluent were analyzed. The samples were collected over a 1/2-hour period at a plant influent flow rate of 5 mgd. Results are summarized in Table 2.
On the basis of a plant flow of 9 mgd (the design flow), partly on the wastewater test data, and partly on educated assumptions, the maximum emission rates of the air pollutants were estimated with and without control. In the case of total VOSs and toxic VOSs, the maximum emissions and risk assessment were based on 100 percent of the substances found in the wastewater being released to the air. In the case of ammonia, CS2, DMS and amines, the emission rates from wastewater to air were estimated by considering the trickling filter as a complete-mix reactor with diffused aeration. When Henry's Law constants could not be found in the literature, calculations were based on the H-constants of similar compounds.
To comply with the relatively tight deadline in the ACO, the consultant was authorized to proceed with design in October of 1991. Construction bids were received on January 17, 1992. Shortly after this date, but before the construction contract was awarded to the low bidder, NJDEP issued the permit to construct. The agency also agreed to revise the ACO project completion date from August 31 to October 31 of that year.
The construction contract was awarded to Remsco Associates, Inc., in April and construction began immediately. The project was completed and the scrubbers put into operation ahead of the October deadline. Construction cost was $1.54 million.
The odor control system consists of air handling and gas deodorization components (Fig 1). Each trickling filter has four fans to exhaust the air from the underdrain, producing a downward air flow at all times (Fig 2). The fans exhaust the off-gases to ducts connected to each scrubber, where the gas is treated before release to the atmosphere through a stack opening 50 ft above grade.
The scrubbing system consists of two scrubber trains connected in parallel, one for each filter. Each consists of two reaction chambers in series, and each of these is 12 ft in diameter and 30 feet tall. Both scrubber trains discharge through the common exhaust stack.
The odorous gases from the filters are contained by a slight vacuum inside the underdrain section, created to induce the downward air flow through the rock media. On each unit the suction from two fans, each rated to 10,000 cfm, is connected to the filter effluent manhole. The suction of two other fans, each rated for 5,000 cfm, is connected to the upstream manhole and to 20 of the filter's vent pipes. The four fans for each filter are connected in parallel and discharge through a common header to the scrubber train.
As described earlier, the consultant had recommended an air handling system capable of withdrawing either 25,000 or 40,000 cfm per filter. This was constructed, therefore the combined exhaust rate through the stack can range between 50,000 and 80,000 cfm when both scrubber trains are in operation.
As odorous gases enter the first reaction chamber, they mix with a fine mist of water, sodium hypochlorite and sodium hydroxide solution. The mist droplets are 5 to 20 microns diam. Scrubbing chemicals are pumped from the storage tanks to the nozzles with positive displacement chemical feed pumps. The fine mist is formed by a nozzle that atomizes the chemical solution with compressed air. A swirling effect induced by the dynamics of the gas flow in the chamber produces provides thorough and intimate contact between the gas and the chemical oxidants. The gas then leaves the first chamber at the bottom and enters the second chamber at the bottom. This provides additional reaction time. Also, a nozzle is located near the bottom of the second chamber. Sodium hydroxide is added at this point to remove any carryover sodium hypochlorite before it reaches the stack. Combined residence time in each pair of reaction chambers is 10 seconds at the 40,000 cfm rate, and 16 seconds at 25,000 cfm.
The once-through scrub solution flows to the bottom of the scrubber chambers and is continuously drained off, along with the reaction byproducts, to the wastewater treatment process. A pH probe installed in the drain line from each scrubber chamber measures the pH of the spent scrub solution. The chemical feed pumps are adjusted manually to maintain the desired pH level and sodium hypochlorite concentration. Chemical consumption is monitored by recording storage tank levels, and solution water flow rate is adjusted manually with rotometers.
Before the treated gases leave the exhaust stack, they pass through a mesh pad mist eliminator near the top of the second reaction chamber to remove entrained droplets. Spray nozzles located under the mist eliminator provide a convenient way to flush the mesh pad with water periodically.
Performance Demonstration Tests
NJDEP required stack testing for emission rates (lb/hr) of the pollutants listed in the permit. Table 4 summarizes the scrubber emissions during the test and the maximum emission rates allowed by the permit. Most of the pollutants were found to be below their detection limit. Only the concentrations of total particulates and hydrogen sulfide were above the analytical limits of detection. In all cases the results proved compliance with NJDEP permit conditions.
In addition, the construction contract required testing for scrubber removal efficiencies of H2S, CS2, DMS and odor units (ED50). The table summarizes the removal efficiencies and the contract requirements, all of which were met.
Performance demonstration test results have shown that the trickling filter emissions have been significantly reduced. As if to confirm this, odor complaints from neighbors have been virtually eliminated.
The off-gases sampled during the pilot study had much higher ED50 values than those measured in the new system's inlet gas stream. Part of the reduction in odor emissions is attributable to the improved ventilation of the filters, and we suspect that the H2S measured in the scrubber inlet is the result of H2S already present in the plant's influent wastewater being released into the air.
Even at the low ED50 levels in their inlets, the scrubbers reduced these levels by half. It should be noted that the accuracy of the odor panel results at low ED50 levels is reduced, partly because background odors are more likely to introduce error, and simply because ranking of the odor intensity by the panelists is more difficult.