Waste Incineration: Heavy Metals
Research Articles
Statistical Analysis of Heavy Metal Data from Municipal Waste Incineration Residues Gerd Bauer, Elisabeth Schachermayer Institute for Water Quality and Waste Management/Waste Management Group, Karlsplatz 13, A-1040 Vienna, Austria
Correspondingauthor: Dipl.-Ing. Dr. Gerd Bauer
Abstract Heavy Metal data from a measurement program at a municipal incinerator in Vienna, Austria are analysed. The experimental design had a nested structure, and analysis of variance is used to break down the total variability into several components; (1) variability between mean hourly concentrations, (2) variability between mean sample concentrations within intervals of an hour and (3) variability due to "measurement" errors. Results of a pre-experiment show, that if grab samples are taken, within hour variation is dominant. The results of analysis of variance were used to calculate an optimal number of samples, which ensure a predetermined accuracy for the mean concentrations. In two main measuring periods with different incinerated waste types, slag and fly ash (ESP dust and boiler ash) data are produced and analysed. For both slag and fly ash, significant differences in measures of location and dispersion between the data sets of the main experiment are found. This proves, that differences in the chemical set up of waste can be detected via incineration residues. Key words: Statistical analysis; experimental design; analysis of variance; waste incineration; heavy metals; slag; fly ash
1
Introduction
Regular analysis of municipal solid waste (MSW) is necessary for several reasons (BRUNNER & ERNST,1986): (1) To identify potentials for recycling or needs for treatment and disposal capacities; (2) to characterise the materials for optimal design of the waste treatment processes, and (3) knowledge of the inputs of waste treatments allows prediction of emissions into the environment. Most measuring campaigns focus on physical composition (e.g. paper, glass, metals), or waste parameters like density or water content. For environmental regulations, knowledge of the chemical composition of waste, and the distribution (transfer coefficients) into environmental compartments is of major concern. The extremely heterogeneous composition of municipal solid waste leads to problems in sample preparation and chemical analysis. As an alternative method, the materials flux analysis a p p r o a c h is suggested (BRUNNER & ERNST, 1986). This method was used several times since the 70's
10
and is based on the mass balance principle (EAWAG, 1982; BERGSTROM &~ LUNDQUIST, 1984; BRUNNER &[ MONCH, 1986, REINMANN, 1989, ANGENEND &[ TRONDT, 1990; SCHACHERMAYER et al., 1995). Analysing the products of waste treatment over a certain period of time, allows the estimation of a mean chemical waste composition in this period. For that purpose, knowledge of statistical influences on concentration data in incineration products is essential to reduce laboratory w o r k and improve experimental design. The objective is to control uncertainty, which means that the uncertainty of a conclusion about the chemical composition of waste, based on a small n u m b e r of measurements at selected points, should be within quantifiable bounds. With regard to indirect observations of the chemical set-up of waste via incineration residues, these aspects have never been the subject of analysis. In this paper, statistical instruments are applied on heavy metal data of municipal waste incineration products, to analyse the variation due to several sources. The effects of sample collection and preparation on the statistical variation are analysed. In addition, data sets representing different periods are compared. The emphasis of the investigation is on aspects of statistical variation, statistical analysis and its relation with experimental design. This is of interest when future measurement programs for determination of heavy metal concentrations in municipal solid waste via incineration products are planned, executed, analysed and interpreted.
2
2.1
Experimental Design Incinerator and Waste Materials
A state of the art municipal solid waste incinerator plant in Vienna, Austria was investigated. The capacity is 270,000 tons of municipal solid waste (waste from household, trade and commerce) per year. The incinerator is equipped with an electrostatic precipitator (ESP), two wet scrubbers and a catalytic denitrification process. Combustion conditions in the furnace are controlled by a digital c o n t r o l system, which adjusts the feeding rate of waste according to its calorific value. The investigations consisted of two parts: A pre-experiment and a main experiment. In the pre-experiment several samples were taken from slag and filter ash (ESP dust and ESPR - Environ. Sci. & Pollut. Res. 3 (1) 10--16 (1996) 9 ecomed publishers, D-86899 Landsberg, Germany
Research Articles bottom ash), waste water and filter cake over a period of three days. The concentration data provided by these samples were used to analyse the variation and to calculate a minimum number of samples, necessary to achieve a predetermined accuracy of the estimated mean values. In the main experiment waste from private households (selected waste from special, non-industrial regions within Vienna) as well as municipal solid waste (regularly selected waste) was incinerated separately. During an incineration period of two days, the same products were sampled and analysed. One goal was to determine the chemical composition of waste from private households and municipal solid waste. These results are presented elsewhere (ScHACHERMAYER et al., 1995). A second goal was to analyse statistical variation of heavy metal data from products of waste incineration as a result of several sources.
Waste Incineration: Heavy Metals
Totalamountof product incinerated ]
Fig. 1: Structure of the sampling procedure in the pre-experiment
I 0 Sampling Points
The heavy metals investigated were: zinc, copper, lead and cadmium in slag and filter ash. Since these elements are concentrated in slag and filter ash (BRUNNER & M(3NCH, 1986), statistical investigations of concentration data in the two measuring periods gives some insight about:
Boiler 9
i[ ~v Wet Scr ubber ~.n:~-~_ r'mo~m_~]
(1) The statistical uncertainty of conclusions based on heavy metal measurements (e. g. indirect estimation of chemical waste composition via products of incineration),
:
Sampling and Analysis
In the pre-experiment, grab samples of approximately 2 kg for slag (number of samples: 36) and 1.5 kg for filter ash (number of samples: 24) were manually taken from the product stream on an hourly basis (6 respectively 4 samples per hour). Slag samples were collected from a conveyor belt (--9 Fig. 2), separated from iron pieces, sieved (
Research Articles
Statistical Analysis of Heavy Metal Data from Municipal Waste Incineration Residues Gerd Bauer, Elisabeth Schachermayer Institute for Water Quality and Waste Management/Waste Management Group, Karlsplatz 13, A-1040 Vienna, Austria
Correspondingauthor: Dipl.-Ing. Dr. Gerd Bauer
Abstract Heavy Metal data from a measurement program at a municipal incinerator in Vienna, Austria are analysed. The experimental design had a nested structure, and analysis of variance is used to break down the total variability into several components; (1) variability between mean hourly concentrations, (2) variability between mean sample concentrations within intervals of an hour and (3) variability due to "measurement" errors. Results of a pre-experiment show, that if grab samples are taken, within hour variation is dominant. The results of analysis of variance were used to calculate an optimal number of samples, which ensure a predetermined accuracy for the mean concentrations. In two main measuring periods with different incinerated waste types, slag and fly ash (ESP dust and boiler ash) data are produced and analysed. For both slag and fly ash, significant differences in measures of location and dispersion between the data sets of the main experiment are found. This proves, that differences in the chemical set up of waste can be detected via incineration residues. Key words: Statistical analysis; experimental design; analysis of variance; waste incineration; heavy metals; slag; fly ash
1
Introduction
Regular analysis of municipal solid waste (MSW) is necessary for several reasons (BRUNNER & ERNST,1986): (1) To identify potentials for recycling or needs for treatment and disposal capacities; (2) to characterise the materials for optimal design of the waste treatment processes, and (3) knowledge of the inputs of waste treatments allows prediction of emissions into the environment. Most measuring campaigns focus on physical composition (e.g. paper, glass, metals), or waste parameters like density or water content. For environmental regulations, knowledge of the chemical composition of waste, and the distribution (transfer coefficients) into environmental compartments is of major concern. The extremely heterogeneous composition of municipal solid waste leads to problems in sample preparation and chemical analysis. As an alternative method, the materials flux analysis a p p r o a c h is suggested (BRUNNER & ERNST, 1986). This method was used several times since the 70's
10
and is based on the mass balance principle (EAWAG, 1982; BERGSTROM &~ LUNDQUIST, 1984; BRUNNER &[ MONCH, 1986, REINMANN, 1989, ANGENEND &[ TRONDT, 1990; SCHACHERMAYER et al., 1995). Analysing the products of waste treatment over a certain period of time, allows the estimation of a mean chemical waste composition in this period. For that purpose, knowledge of statistical influences on concentration data in incineration products is essential to reduce laboratory w o r k and improve experimental design. The objective is to control uncertainty, which means that the uncertainty of a conclusion about the chemical composition of waste, based on a small n u m b e r of measurements at selected points, should be within quantifiable bounds. With regard to indirect observations of the chemical set-up of waste via incineration residues, these aspects have never been the subject of analysis. In this paper, statistical instruments are applied on heavy metal data of municipal waste incineration products, to analyse the variation due to several sources. The effects of sample collection and preparation on the statistical variation are analysed. In addition, data sets representing different periods are compared. The emphasis of the investigation is on aspects of statistical variation, statistical analysis and its relation with experimental design. This is of interest when future measurement programs for determination of heavy metal concentrations in municipal solid waste via incineration products are planned, executed, analysed and interpreted.
2
2.1
Experimental Design Incinerator and Waste Materials
A state of the art municipal solid waste incinerator plant in Vienna, Austria was investigated. The capacity is 270,000 tons of municipal solid waste (waste from household, trade and commerce) per year. The incinerator is equipped with an electrostatic precipitator (ESP), two wet scrubbers and a catalytic denitrification process. Combustion conditions in the furnace are controlled by a digital c o n t r o l system, which adjusts the feeding rate of waste according to its calorific value. The investigations consisted of two parts: A pre-experiment and a main experiment. In the pre-experiment several samples were taken from slag and filter ash (ESP dust and ESPR - Environ. Sci. & Pollut. Res. 3 (1) 10--16 (1996) 9 ecomed publishers, D-86899 Landsberg, Germany
Research Articles bottom ash), waste water and filter cake over a period of three days. The concentration data provided by these samples were used to analyse the variation and to calculate a minimum number of samples, necessary to achieve a predetermined accuracy of the estimated mean values. In the main experiment waste from private households (selected waste from special, non-industrial regions within Vienna) as well as municipal solid waste (regularly selected waste) was incinerated separately. During an incineration period of two days, the same products were sampled and analysed. One goal was to determine the chemical composition of waste from private households and municipal solid waste. These results are presented elsewhere (ScHACHERMAYER et al., 1995). A second goal was to analyse statistical variation of heavy metal data from products of waste incineration as a result of several sources.
Waste Incineration: Heavy Metals
Totalamountof product incinerated ]
Fig. 1: Structure of the sampling procedure in the pre-experiment
I 0 Sampling Points
The heavy metals investigated were: zinc, copper, lead and cadmium in slag and filter ash. Since these elements are concentrated in slag and filter ash (BRUNNER & M(3NCH, 1986), statistical investigations of concentration data in the two measuring periods gives some insight about:
Boiler 9
i[ ~v Wet Scr ubber ~.n:~-~_ r'mo~m_~]
(1) The statistical uncertainty of conclusions based on heavy metal measurements (e. g. indirect estimation of chemical waste composition via products of incineration),
:
Sampling and Analysis
In the pre-experiment, grab samples of approximately 2 kg for slag (number of samples: 36) and 1.5 kg for filter ash (number of samples: 24) were manually taken from the product stream on an hourly basis (6 respectively 4 samples per hour). Slag samples were collected from a conveyor belt (--9 Fig. 2), separated from iron pieces, sieved (
