AMMONIA AND HYDROGEN EMISSIONSFROM AN INDUSTRIAL WASTELANDFILL: A CASE STUDYJ. FELIUBADALÓEntitat Metropolitana de Serveis Hidràulics i Tractament de Residus, C/ 62,nº 16 - 18, Ed. B, 08040 Barcelona, Spain.SUMMARY: Emissions from IW of gases other than methane an carbon dioxide are until now notstudied as in MSW landfills are. This paper deals with the origin, monitoring, temporal evolution andmathematical modelization of ammonia and hydrogen emissions from a IW landfill. Melting salts from arecycled aluminium metallurgical process, among some other kinds of waste, compose the landfilledmass.1. INTRODUCTIONEmissions of MSW landfills are currently a well-known subject, in terms of quality (range of major:methane and carbon dioxide, and minor compounds) and in terms of quantity, rate and temporalevolution of these parameters. Related to mono-industrial waste landfills, a lack of data is found, perhaps due to the relativescarceness of this kind of facilities and also to the extreme diversity of industrial substances disposed inthem. Because of a number of reasons (Relea et al., 1995), the Authorities in waste management of theMetropolitan Area of Barcelona have been applying historically an splitting strategy between wastes ofdiverse origins. This strategy is currently being incorporated into Spanish and European regulations onwaste landfilling. Moreover, these Authorities have played a role of substitution of private initiative when the industrialwaste management was not technically and economically attractive for it. In this context, at the last1980s, the EMSHTR established and operated for about seven years an industrial waste landfill in aancient clay quarry at Cerdanyola (a village about 15 Km at the North-West from Barcelona). Theextremely low values of hydraulic conductivity of the clay (even below 10-11 m.s-1) made it speciallysuitable for that purpose. This landfill was located about 2 Km at the East from that referred by thesame authors (op. cit.). In principle, the landfill was established for a single industrial waste: the melting salts from recycledaluminium metallurgical process. Further, some other kinds of waste were also disposed there, namelynon-special general industrial waste, industrial sewage sludge from the treatment of
brine destined to electrolytic production of chlorine and caustic soda, and polluted soils from an ancientchemicals plant at a nearly Barcelona borough. The approximate amounts and percentages of every kind of waste are summarized in Table 1.Table 1 - Amounts and kinds of landfilled waste Kind of waste Amount, t Percentage Melting salts 118.000 18,4 Industrial sewage sludge 77.000 12,0 Polluted soils 7.000 1,2 Non-special general industrial waste 440.000 68,5 Total 642.000 100,02. ORIGIN AND CHEMICAL FEATURES OF MELTING SALTSThe metallurgical process mentioned above lies basically in the melting, in a chemically reducingatmosphere, of a mixture of aluminium recycled goods and melting salts (almost all sodium andpotassium natural chlorides). Molten salts float over the molten aluminiun, protecting it from oxydationand absorbing in its mass most of the aluminium impurities. As a result of this process, and specially due to the reducing character of the atmosphere in which ittakes place, in the residual salts is present a certain amount (2 - 3%) of metallic aluminium, besides withsome other by-products: aluminium nitrides, hydrides and sulphides, formed by the reaction of thiselement with nitrogen and hydrogen, both present at the reducing atmosphere, and by the reaction of itand of salts alkaline metals with the sulphur compounds usually present at the fuel used for the meltingprocess. The overall chemical reactions describing these processes can be summarized, respectively, asfollows: 2 Al + N2 ???> 2 AlN 2 Al + 3 H2 ???> 2 AlH3 + 3 S + 2 Al ???> S3Al2 and S + 2 Na ???> SNa2If waste salts are put in contact with water, these componds react with it, giving aluminium and sodiumhydroxides and, respectively, ammonia, hydrogen and hydrogen sulphide, probably according to thefollowing reactions: AlN + 3 H2O ???> NH3 + Al(OH)3 AlH3 + 3 H2O ???> 3 H2 + Al(OH)3 S3Al2 + 6 H2O <???> 3SH2 + 2Al(OH)3, and SNa2 + 2 H2O<???> SH2 + 2 NaOH
As it can be seen, in all cases these processes give some alkaline or amphoteric solid compoundsbesides with three kinds of gases, all of them with a negative environmental impact due to variousreasons: toxicity and corrosivity in the case of ammonia, explosiveness in that of hydrogen, and toxicity in that of hydrogen sulphide. By the way, it is worth to point up that the alkalinity of sewage sludge disposed together with the casting salts contributes to the stabilization of the alkaline compunds formed. In the other hand, levels of hydrogen sulphide were found neglectable when compared with those of ammonia and hydrogen. Therefore, it was decided to monitorize only the last two, whose levels were in the range of percentages, whereas that of hydrogen sulphide was in the range of mg/m3. 3. SAMPLING AND ANALYSIS METHODS AND PERIODICITY OF SAMPLING 3.1. Sampling and analysis methods Ammonia, because of its chemical properties, which made impossible its instrumental analysis, was collected and analyzed by a “classical” method: chemical absorption in sulphuric acid and colorimetric dosing by Nessler method. Sampling was performed by means a "train" of three maxi-impigers, the first two filled with H2SO4 0,1N and the third, empty, acting as a trap. Gas to be sampled was forced trough the train by a peristaltic air pump. Hydrogen was analyzed at a laboratory by an instrumental method (gas cromatograpy). So, its sampling lay just in the collection in an hermetic poliethylene bag by means of an air pump. 3.3. Periodicity of samplings and unit of time It has been performed a total of 22 samplings and subsequent analysis, distributed about evenly in time, between February 1995 (just after the cessation of landfilling), and March 1998, when concentrations of both gases dropped below worrying levels. Although intervals between samplings are not exactly equal, in the following they will be considered as constant, in order to simplify calculations and graphics. Consequently, the time unit used will be that interval, and calculations will be done on that basis. An elemental calculation gives that the equivalence factor with year is 6,6316 samplings per year. 4. RESULTS Results of the entire series of ammonia and hydrogen analysis are presented together at Figure 1, in order to highlight the appoximately parallel temporal evolution of concentrations of both gases (note that those of hydrogen have been reduced by a factor of 10). The most appealing feature of both evolutions is, undoubtedly, the abrupt rising they show at the 11th sampling. As it will be discussed further, both evolutions could be approached quite accurately by a first order decay mathematical model, but that describing the evolution until that point fails to be valid from it. However, the remaining part of both evolutions can be described with about the same accuracy as well by another first-order decay equation or, more precisely, by an equation of the same kind that the first but with other parameters.
As it can be seen in Figure 1, the degree of parallelism of the evolutions of ammonia and hydrogen allows the description of both phenomena by the same kind of mathematical model, as explained at Chapter 6. 0,8 Concentrations, % v/v 0,7 0,6 0,5 Ammonia conc. 0,4 Hydrogen conc./10 0,3 0,2 0,1 0 0 2 4 6 8 10 12 14 16 18 20 Sampling nºFigure 1. Evolution of ammonia and hydrogen concentrations by volume.5. DISCUSSION OF RESULTSLooking for an explanation for the abrupt rising of concentrations described above, and havingconsidered that the rising is not attributable to any landfilling of materials (since it had completely ceasedbefore the beginning of sampling) nor to ambient temperature (since the rising has not a seasonal -depending pattern), the sole one that appeared to be reasonable was the dependence of waterinfiltration. In spite of the quality of landfill capping performed (1 m. of clay, 0,20 m. of draining inertmaterial and 0,80 m. of topsoil), it appears obvious that any amount of rainfall and runoff water canreach the melting salts and re-enhance the production of ammonia and hydrogen accordingly toprocesses described at Chapter 2. This assumption about water intrusion into landfill body s corroborated by the fact that even at ipresent, that is, four years after the closure, there is still a certain production of leachate. Moreover, the parallelism of evolutions of ammonia and hydrogen pointed at Chapter 4, stronglysuggests a common cause for the rising of both concentrations. So, in order to investigate the influence of water intrusion, it has being plotted (see Figure 2) theevolution of rainfall (directly measured in place) during the sampling period. The comparison of itsprofile with that of concentrations shows that:• It occurs a first strong peak of rainfall between samplings 6 and 10, with a total of 368 l/m2, whereas the rising of concentrations occurs at sampling 11, that is, a few months later.• A second important peak of rainfall occurs between samplings 17 and 20, with a total of 382 l/m2. In opposition to the precedent, this peak has no correspondence with any one of concentration evolution.These facts would be consistent with the explanation of, whereas at the time af first rainfall peak therewas still any amount of aluminium hydrides and nitrides to react with water and so produce enough
ammonia and hydrogen to give a noticeable peak in its evolution, by the time of second rainfall peak,these substances were already exhausted, and thus, it does not occur a second evolution peak. 140 120 Rainfall, l/m2 100 80 60 40 20 0 0 2 4 6 8 10 12 14 16 18 20 Sampling nº Figure 2. Evolution of rainfall. 6. MATHEMATICAL MODELS 6.1. General Like in municipal organic solid waste (Coops et al, 1995), the kinetics of the reaction of aluminium nitrides and hydrides with water to give aluminium hydroxide and, respectively, ammonia and hydrogen can be described, whit more or less accuracy, by mathematical models based on equations depending on the kinetic order of reactions taking place. Since in the case in study reactions are at least bimolecular, the models should be of second or even further orders (Babor and Ibarz, 1962). Nevertheless, second and further orders models are very difficult to apply in practice, and so its practical usefulness is limited; so, in this case it was taken a first order model, that is, one giving a negative exponential profile for the reaction rate or, correspondingly, for concentrations of both gases. First order models are, as it is well-known, based on the following general equation: -kt Ct = C0* e (1) where Ct is the concentration at the instant t, C0 that of the initial instant and k is the velocity constant of the process. The last is related with Ct and C0 by the equation k = (1/t) * ln (C 0/Ct ) (2) Thus, to know C0 and Ct gives the value of k or, in other words, completely determines the equation or equations ruling the mathematical model. Therefore, for the complete determination of a first order
equation, it is sufficient to have the initial value of concentration (C0) and its value after a known time t (C t ). It is easy to demonstrate that equation (2) can be generalized to any pair of values of t: k = 1/(t2-t1) * ln (C1/C2) (3)If, as in this case, one looks for the modelization of a set of empirical data, these values of C might bechosen so as to positive and negative errors of model were approximately compensated, in order tomaximize the accuracy of it. This can require a number of trials. Moreover, in this case, the special feature of the set of values reported in Chapter 5 leads to theneed of two equations, the first describing the evolution before the secondary peak reported there andthe second accounting for the remainder of the obtained values. It is worth to point here that thisprocedure has an evident physical interpretation. In effect, the overall phenomenon can be described interms of the succession of two approximately exponential decay processes, the second of them startingfrom a secondary concentration peak due to an external incidental cause.6.2 Calculation of coefficientsThe coefficients of equations constituting the mathematical model [(2) and (3) respectively for the leftand right part of it, as it has been explained above] have been derived, both for the cases of ammoniaand hydrogen, from empirical data plotted at Figure 1.The values used for calculation of parameters for ammonia and hydrogen models are those listed inTable 2:Table 2 - Values of t and C used for calculation of ammonia model parameters Ammonia Hydrogen Instant Concentration, % Instant Concentration, % t0 = 0 C0 = 0,8 t0 = 0 C0 = 7,8 t1 = 6,8 C1 = 0,2 t1 = 10 C1 = 0,5 t2 = 11 C2 = 0,45 t2 = 11 C2 = 4,5 t3 = 15,5 C3 = 0,07 t3 = 15,5 C3 = 0,7The application of the corresponding values to expressions (2), (3) and (1) gives for the wholemathematical model for ammonia the following equations: Cat1 = 0,80 * e -0,2038 t (4) and C a t2 = 0,45 * e -0,4134 t (5)Being (4) valid between instants 0 and 10, and (5) for the remaining period.And the equations fot hydrogen model are Cht1 = 7,8 * e -0,2747 t (6) and C h t2 = 4,5 * e -0,4134 t (7)
Being as well (6) valid until instant 10, and (7) between 11 and 21ones.It is to be noted that, accordingly with the last statement of paragraph 4.1, the exponent of equation (7)is the exactly the same of (5) one, and that its coefficient is that of (5) multiplied by a factor of 10. Note as well that the unit of all exponents is the (sampling intervall)-1. To express them in year—1they must be divided by the factor 6.6316, as stated at 4.3. 6.3. Comparison between observed and model-derived values of concentration. In order to show graphically this comparison, observed and model - derived values have been plotted together, respectively for ammonia and hydrogen, at Figures 3 and 4. 7. CONCLUSIONS • As it could be expected, the decay of concentrations for the two studied gases (both of them coming from inorganic substances) is much faster (by a factor of about 10) than that of methane and carbon dioxide in biogas, due, instead, to the decomposition of organic products. • The rate of generation for ammonia and hydrogen appears to be strongly related to water intrusion into waste mass. This statement is supported by the observational evidence and by its explanation in chemical theoretical terms. No other external factors appear to have a comparable influence on generation rate. • In spite of deviations due to external causes other than water intrusion, first order mathematical model appears to be a quite good approach to observational results. In the other hand, the enhancements of generation rate due to water intrusions can also be approached just re-adjusting the parameters of model to observational data. • If these enhancements do not occur, the simplicity of the first order model allows to derive its parameters (that is, to define completely the model equation) from just two values of concentration at two known instants of time. ACKNOWLEGEMENTS
Author gratefully acknowledges Adoración Pascual, Maria Gràcia Rosell, Xavier Guardino and Emili Castejón, from the Instituto Nacional de Seguridad e Higiene en el Trabajo, at Ammonia concentrations, % v/v 0,8 0,7 0,6 0,5 Observed conc. 0,4 Math. model conc. 0,3 0,2 0,1 0 0 2 4 6 8 10 12 14 16 18 20 Sampling nº Figure 3. Ammonia observed versus mathematical model derived concentrations. Hydrogen concentrations, % v/v 8 7 6 5 Observed conc. 4 Math. model conc. 3 2 1 0 0 2 4 6 8 10 12 14 16 18 20 Sampling nºFigure 4. Hydrogen observed versus mathematical model derived concentrations.whose laboratory in Barcelona has been performed all the analytical work, and Miquel Gelabert, JuanLeyva and Josep Mª Biescas, from the EMSHTR, who have shared with Mrs. Pascual and himself thesampling works.REFERENCESRelea F., Feliubadaló J. and Montells R. MSW and NSIW landfilling: an emission comparison. Proceedings Sardinia 95, Fifth International Landfill Symposium, CISA publisher, Cagliari, vol. III, 223-234.Babor J.A. and Ibarz J. Química General Moderna, Editorial Marín, Barcelona 1962, 294-295.
Coops O., Lunning L., Oonk H. and Weenk A. Validation of gas formation models. Proceedings Sardinia 95, Fifth International Landfill Symposium, CISA publisher, Cagliari, vol. I, 634-646.