|
Part 4.4 Abatement of releases Criteria for abatement There was little regulatory interest in releases of MDI or TDI until the mid 1990s. This was primarily because of the low levels of releases: also, there had been little regulatory activity associated with chemicals, in general. Further, it had been well established that MDI and TDI have little environmental impact and that they decay in air. However, several regulatory authorities have now established ex stack maximum release concentrations for these materials. Further, in the US there are fenceline (community) maximum concentrations for MDI and TDI, within the IRIS programme of establishing community risk from releases of chemicals. Ex stack concentrations are relatively easy to determine, using well-established methods. Fenceline compliance measurements are more complex: they involve direct measurements at the limits of analytical methodology (typically 1 to 100 parts per trillion), or the use of modelling of ex stack concentrations. Modelling involves the use of meteorological conditions at the site, factory site parameters and geographical features of the surrounding area. The topic is discussed and various models explained in Part 3 section (iii) of this Toolbox. Abatement methods Several methods have been investigated to reduce releases of MDI or TDI present in the exhaust air of ventilation stacks. Of the methods tested, only neutral and alkali aqueous scrubbing and carbon adsorption methods have proved to be both practically and economically viable. The choice of method depends upon the co-releases of chemicals which also have to be abated. Pilot studies have been carried out on the destruction of releases using OH radicals or biological media but these have not led to commercial development of the methods, probably due to the simplicity and effectiveness of carbon or aqueous systems. Activated carbon adsorption systems Design of activated carbon units A wide range of designs has been tested, and selected ones used commercially, over the past 25 years. The carbon has been used as plates or in granular form. Whilst carbon in the plate form avoids the problem of air channels forming, the problems of sealing the plates in the exhaust ducting have been found to be significant. Carbon, as granules, has been used in systems with multiple layers, or as a single large bank. A problem with multiple layers is to prevent extensive channelling (and hence short contact time of the carbon with the exhaust gases) at each layer. Modern systems use large banks of granular carbon either in a cylinder or in an annular configuration. There are five critical aspects to be considered:
The contact time of the exhaust air with the carbon is defined as the velocity of air flow through the bed divided by the bed thickness: for example, a linear velocity of 0.4 m/sec (1.3 ft/sec) and a bed thickness of 1 metre (3.3 ft) would give a contact time of 2.5 seconds. Control of linear velocity, and hence residence time, of the exhaust gases in the adsorption unit is critical. For effective TDI removal one manufacturer has recommended that the air velocity should not exceed 0.5 m/sec (1.6 ft/sec) and that the contact time should be at least two seconds. Adsorbers using these criteria have operated successfully for over 15 years, reducing TDI emissions to acceptable discharge concentrations. Since beds may use several tonnes of carbon and last for years, the choice of carbon type is critical. It is important that experts be consulted and, if necessary, small-scale laboratory tests carried out. It is not difficult to calculate the required contact time. However, in that respect it should not be assumed that there is even flow through the carbon. Channelling in the carbon bed, which results in air passing preferentially through open channels formed mechanically in the carbon bed, leads to reduced contact time of the exhaust gases with the carbon, and hence to lower adsorption efficiency. Clogging of the carbon bed, where carbon particles adhere to each other at the inlet of the bed to form a mass of low permeability, can also be a problem: the pressure across the bed becomes so high that exhaust gases can not be forced through it. Mechanical break-up of the carbon at the inlet end of the bed may be necessary to prevent this. An associated phenomenon is blocking of the carbon by dust and other debris. Some manufacturers use a so-called sacrifice layer before the carbon adsorption unit. This is a unit containing cut scrap flexible polyurethane foam or other removable material. This has the function of reducing channelling of the air by preventing any debris from the air collecting at the inlet of the carbon adsorption unit. Further, it is well known that flexible foam adsorbs diisocyanates very strongly, so the concentration of diisocyanate entering the carbon adsorption unit is already reduced. The sacrifice layer is replaced very frequently at low cost. This use of flexible foam for the abatement of diisocyanate has been described (Wood et al., 1993). The concentration of TDI decreases as the air passes along ducting to the carbon bed, due to adsorption on the side walls. One company using a substantial length of ducting found a reduction of 30% to 40% of the initial TDI concentration. Thus, when measuring the efficiency of a scrubber unit the inlet concentration should be measured near to the entrance of the scrubber. There has been considerable experience with carbon beds, some of which is described below. In one case, the carbon bed was changed after seven years operation, not because of loss of efficiency (still >99%), but because of the increase in back pressure. The use of a simple acrylic filter impregnated with activated carbon before the main carbon bed, not only removed particulates, but also reduced the concentration of TDI by approximately 20%. The acrylic filter needed to be renewed frequently. In another case, the carbon bed was reported as freezing at very low ambient temperatures and it consequently failed to operate until heated by warm air. In Table 20 are given the results of analysis of the same type of carbon used in three different carbon adsorbers over a period of seven years (III unpublished data). It had already been ascertained that the maximum loading of the given carbon was about 25g of reaction product per 100g of carbon calculated as TDI product. It can be seen that there was a gradation of loading from the inlet to the outlet in all three adsorbers. In Adsorber 1 the saturation loading (25.4 g/100 g carbon) had been reached and the remaining lifetime of this adsorber was very limited because of the reducing carbon contact time: in the cases of Adsorbers 2 and 3 there was still considerable capacity, this being due to the lower concentrations of TDI being passed through them. Table 20: Loadings of TDI reaction product on carbon adsorbers
Analysis of used carbon Samples of activated carbon from the above three carbon adsorbers were examined. The spent carbon was grey in colour. Analysis of this spent carbon showed that neither TDI nor TDA was detectable (detection limit: 5 mg/kg) so that any nitrogen present would probably be as polyureas. The results of chemical analysis are given in Table 21, where the figures are expressed as percentages of the carbon in the bed (w/w). In addition to the results of analysis are given calculated values of the percentage of polyurea which would be formed if all the nitrogen were converted to polyurea: the value approaches that of 25 g/100 g carbon given in Table 20. Table 21: Analysis of carbon from beds
Reactivation of activated carbon The above carbon was reactivated by heating in a process used by the original supplier. The results of reactivation tests showed that the grey colour and polymerization products could be totally removed along with the adsorbed impurities. The reactivated product was found to have good adsorption capacity, as measured by benzene adsorption, which is used as a standard. It was concluded that the spent carbon could be reactivated and re-used in air purification applications. Aqueous scrubbing Aqueous alkali scrubbing of MDI and TDI has been used with success in the UK for more than 20 years. Typical systems use a bed packed with polypropylene spheres to provide a large surface area for the scrubbing agent, sodium hydroxide solution. The alkali trickles through the bed under gravity, the extract air is forced upwards and permeates the bed, and the diisocyanate is removed by reaction with the alkali. The alkali is recirculated. Grey and Chadwick (1979) published details of the design of the Cleme Gas Scrubber, and reported that a conventional twin impingement plate scrubber unit gave only approximately 50% efficiency. Modifications were carried out to the design and eventually an efficiency of greater than 90% was achieved. It should be noted that the concentration of diisocyanate in the gas phase before abatement is so low that there is negligible change of the temperature in the aqueous scrubber due to the heat of reaction. Grey and Chadwick reported that the first unit installed made no provision for easily monitoring and adjusting the strength of the sodium hydroxide scrubbing liquor. As a result of the removal of carbon dioxide from the exhaust ventilation gas stream, sodium hydroxide became progressively consumed and converted through sodium carbonate to sodium bicarbonate. The scrubbing action of 0.3% to 5% sodium carbonate solution was known to be as efficient as 0.25% to 4% sodium hydroxide, but sodium bicarbonate was less effective. Steps were taken to improve the control of alkalinity of the scrubbing medium using an automatic dosing system, which replenished the sodium hydroxide level by injecting 30% aqueous sodium hydroxide through appropriate valving and pumps. Other designs of aqueous scrubber have been developed. One foam manufacturer evaluated a packed tower design. The work indicated that an efficiency of greater than 95% using alkali scrubbing could be guaranteed. Efficiencies of the order of only 80% were achieved when water rather than aqueous alkali was used as the scrubber liquor. A possible improvement to this problem was used by one foam manufacturer, who found that the replacement of alkali by aqueous solutions of urea gave scrubbing efficiencies close to those obtained with alkali, and significantly higher than those with water. An improvement over water scrubbing might be to combine it with aqueous oxidation. It is already known that MDI and TDI are oxidized by hydroxyl radicals which could be generated in the aqueous system. Pilot trials have indicated that such a system would be viable (Barker and Jones, 1988) Neutral aqueous scrubbing can be achieved by the use of venturi systems. Hurd (1988) has described the use of single stage and multi-stage venturi installations on two UK flexible foam slabstock plants. A single venturi installation on one plant operating at approximately 50,000 cubic feet per minute (cfm) of air gave about 50% efficiency with inlet TDI concentrations of around 1000 ppb. Very good results have been obtained, however, using three venturi scrubbers in series. This plant operated with an air extraction rate of 64,000 cfm (109,000 m3/h) of air. The inlet TDI concentration was monitored continuously by a paper tape monitor. The data from the monitor were fed to a microprocessor. Automatically, the number of venturis required to reduce the outlet TDI concentration to below 20 ppb were brought into operation, assuming 55% efficiency in each successive venturi jet in the series of three jets. The full series of three were required only when particular foam formulations leading to higher levels of TDI emissions were being run. A maximum scrubbing efficiency of not less 90% was obtained with all the venturis in operation. Water usage was high, however, so a recirculation system was established. Effluent from aqueous scrubbers There are widespread regulations regarding the release of aqueous factory waste into water systems. In order to conform to acceptable standards a knowledge of the composition of the scrubbing liquors, and the local criteria for dilution and disposal, are essential. The effluent waters from two polyurethane factories releasing effluents into water systems were analyzed within an III programme (Chapman, 2000). One of the factories used a water scrubber, whereas the other factory employed an alkali scrubber. TDI could not be detected in either effluent. However, low levels of its reaction product, toluene diamine (TDA), were found in both. TDA is expected to be easily removed from the effluent by a water treatment plant since it is very strongly adsorbed onto the activated sludge (Cowen et al., 1998). Measurements of Total Organic Carbon (TOC), Dissolved Organic Carbon (DOC) and Chemical Oxygen Demand (COD), along with concentrations of various chemical species were reported. The overall conclusion was that upon natural dilution of the effluent, it would have similar characteristics to those of water from public treatment plants. Solids formed in the neutralization process, and removed from the above aqueous effluent, should be disposed of in conformity to local regulations. Whilst there has been a general trend to use carbon abatement rather than the less efficient aqueous systems, interest in aqueous systems continues (Griggs et al., 2000). Efficiencies of TDI abatement: pilot studies Urano at Yokohama National University (1978) carried out work on the use of activated carbon to adsorb TDI. He reported that an adsorption capacity of 0.3 to 0.5 g TDI/g carbon would be achievable even at low concentrations of TDI in air. These values should be viewed in the light of those given in Table 20. A very high adsorption rate for CFC-11 (which was then used universally as a blowing agent) was also reported. He also reported that for triethylenediamine (TEDA), a widely used polyurethane catalyst, the adsorption capacity of the carbon was large, but the adsorption rate was low (this may have been due to adsorbed TEDA being lost from the carbon to the exhaust stream). An important contribution to the data on the use of activated carbon for TDI removal and CFC recovery was made by Gans et al. and by Sporon-Fiedler. Gans and his co-workers at Stuttgart University carried out a detailed investigation of the removal of TDI and CFC recovery from polyurethane emissions, both on the laboratory scale and on a large scale pilot plant, in collaboration with industry. Further, more detailed analytical tests on this pilot plant were reported by Nutt and Skidmore. They reported that the unit was >99% efficient for TDI removal and that CFC-11 recovery was also viable. References on carbon abatement Efficiency of TDI abatement: large scale experience using activated carbon units Following the pilot scale tests reported above, a number of carbon adsorbers were installed on flexible polyurethane foam plants, notably in the UK, Holland and Belgium. Data are now available from these units in terms of efficiency, life and the cost per tonne of TDI treated. Two units which had been operating for over seven years showed no detectable TDI at the exit from the scrubbers. It has been found that a reduction in the air volumes used in polyurethane flexible foam slabstock production was necessary to obtain efficient and economic abatement methods, since capital costs, running costs and space requirement of abatement equipment all increase substantially with increased air volume. Costs will also of course be dependent upon plant output, space requirements and other factors. Substantial reductions of air volumes can be achieved by a study of operating and ventilation conditions. One company, when installing carbon adsorption units, reduced exhaust volumes on a flexible foam slabstock line from approximately 80,000 m3/h to 45,000 m3/h using extraction above the side wall instead of the conventional overhead exhaust. In spite of this change, concentrations of TDI in the workplace did not increase. The results from a later research programme confirmed the efficiency of activated carbon units for TDI removal (see Table 22 for example). Running costs vary depending on the variables outlined but, as a general guide, are of the order of US $2.5 per tonne of TDI processed by the factory (2001). Table 22: The efficiency of carbon bed abatement of TDI from flexible foam production
Abatement of releases from flame lamination In Table 23 are given the effects on concentrations of MDI and TDI releases from flame lamination of MDI-based and TDI-based polyurethane foams of using carbon abatement (Glover and Maddison, 2000). Table 23: The abatement of MDI and TDI from flame lamination processes
All concentrations of MDI or TDI, post abatement, were below any regulatory maximum requirement, and below the detection limits of the method, which were 0.03 mg/m3 for MDI and 0.02 mg/m3 for TDI. The authors also reported on the abatement of carbon dioxide, carbon monoxide, nitrogen oxides, hydrochloric acid and hydrocyanic acid from the exhaust stream. In most cases gases were reduced to, or near to, the level of the analytical detection limits. This work was carried out using new carbon, so the above values reflect optimal abatement for that plant. However, since that study was completed (in 1993) there have been significant developments in the treatment of carbon to adsorb specified chemicals such as hydrochloric acid, hydrocyanic acid and amines, in the last case using acid-treated carbon. Abatement of blowing agents, amine catalysts and other species It is beyond the scope of this text to deal with the abatement of species other than MDI and TDI. Furthermore, individual abatement scenarios need to be examined to allow meaningful approaches to co-abatement to be made. The citations provided below give a considerable amount of information about the concentrations of chemical species which may be released from MDI- or TDI-based processes. References to studies about diisocyanate co-releases In these reported studies, co-releases were determined, in most cases before and after abatement. Flexible foam slabstock
Flexible foam moulding
Flame lamination
Rigid foam slabstock and boardstock
MDI and TDI: summary of release and abatement data In Table 24 are summarized the data detailed in tables above. Table 24: Summary of pre-abatement releases
In Table 25 are given data collected on the abatement of TDI emissions. There are no MDI abatement data available because the unabated releases were lower or much lower than the prevailing regulatory limits. Table 25: Summary of TDI abatement data (carbon beds)
The values given in Table 24 are drawn from the studies cited earlier in this Part 4 of the Toolbox. The ranges of results are probably representative of the global scene, given that the studies were of the applications consuming most of the MDI and TDI used worldwide, and given that the same proprietary manufacturing processes for these applications are found worldwide. The key parameter is mass loss of diisocyanate per mass of diisocyanate processed, based on a per annum calculation, whether as g/tonne or % diisocyanate lost. Ex stack concentration is a less fundamental parameter, being inversely proportional to the volume flow of exhaust ventilation air, to an approximation. Releases of MDI range from <6 x 10-7% to 6 x 10-4% of total MDI throughput, which equate to a range of <0.8 mg/tonne to 2 g/tonne, in the very high volume rigid boardstock and rigid foam slabstock sectors. Losses of TDI range from 2.5 x 10-3% to 9 x 10-3%, which equate to the range of 1 g/tonne to 90 g/tonne TDI, for the dominant TDI usage sectors of flexible foam slabstock and moulding. The unabated concentration and annual loss values for MDI were extremely low, and below all ex stack limits which applied, the most stringent of which was (and still is in 2002) 0.1 mg NCO group/m3 (equivalent to 0.3 mg MDI/m3) as required by the UK Environment Agency. Accordingly, the values calculated for the efficiency of abatement (Table 25) relate to TDI or TDI-based polyurethane (flame lamination) only.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||