PRESENTED AT: ISA AD 2011 Analysis Division Symposium League City, TX

Keywords

Sulfur Dioxide, Direct Extractive, Emissions Monitoring, CEMS, Ultraviolet

Abstract

Sulfur Dioxide is classed by the Environmental Protection Agency (EPA) and all state environmental authorities as a priority pollutant and is commonly present in stack gas emissions from a multitude of sources including refining, chemical processing, power generation and mining. Permitted levels for sulfur dioxide emissions have continuously been reduced over time, and current new construction regulations in several districts require emissions levels in the low parts per million levels.

Continuous emissions monitoring by absorbance spectroscopy has long been a mainstay of sulfur dioxide analysis. However, the reduced emissions levels being promulgated in new rulings are placing increasing demands for high sensitivity and baseline stability. A newly developed ultraviolet absorption spectrometer has been applied to this difficult measurement, and results are presented herein.

Indroduction

Ultraviolet Absorption Spectroscopy has long been one of the prominent methods for measurement of sulfur dioxide in stack gas applications. The method is relatively interference-free, reliable and robust. This has contributed to it being the favored instrumental method used by source testers during Relative Accuracy Test Audits (RATA).

In the last decade, environmental regulations on emissions have become increasingly rigid, and allowed emissions levels for Sulfur Dioxide have continued to drop, such that many new permits require stack gas concentrations in the single-digit ppm range. Many direct extractive CEMS systems are not capable of measuring sulfur dioxide with sufficient accuracy at the low levels required for new permits. In addition, the sample systems for such measurements must to be optimized to minimize bias.

The Requirement

Federal environmental regulations are of course mandated by the US EPA. As such, there are specific and well-defined emission limits for various industries such as refineries, sulfuric acid plans, cement plants and petrochemical facilities. Typically, regulations such as CFR 40 Part 60 SubPart J (refineries) or SubPart H (sulfuric acid plants) result in allowed emissions of sulfur dioxide in the 250 ppm range from many operating facilities.

State regulations use the EPA regulations as guidelines, but may be more stringent. Historically, California has often been the leader in such stringent regulations and the implementation there often indicates the future course of other states. While SubPart J allowed refinery SRU emissions of 250 ppm, the majority of refineries in the Los Angeles Basin (and under the jurisdiction of the South Coast Air Quality Management District [SCAQMD] ) have been operating in the 100 ppm range. The Reclaim Act (RULE 2002. ALLOCATIONS FOR OXIDES OF NITROGEN AND OXIDES OF SULFUR) in California was initially adopted in 1994 and has now gone through its third revision. The revised ruling further reduces sulfur dioxide and nitrogen oxides emissions levels and will be implemented as of 2012. The rule is based on the capabilities of the Best Achievable Retrofit Control Technology (BARCT). The new standards for emissions of sulfur dioxide are as follows:

TABLE 1 – RECLAIM SOX TIER III EMISSION STANDARDS

These dramatic reductions in emission rates (in many case 95%) will require upgrades to the emissions monitoring systems in place, and many emissions analyzers currently in place will need to be replaced with lower range units.

The Analyzer

The spectrometer used was of a dual-beam, multi-wavelength design, as shown schematically in Figure 1. The light sources used in this device are hollow cathode lamps, chosen specifically for their narrow emission lines at specific and reproducible wavelengths. The measure and reference detectors are photomultiplier tubes (PMT), chosen specifically for their excellent photometric accuracy and linearity [1]. Fixed filters in front of each light source are used to isolate specific emission lines.

FIGURE 1 – M9000 SERIES SPECTROMETER

The dual-beam design was chosen to provide exceptional baseline stability. In this device, a separate reference and measure path is maintained for each measurement wavelength employed. The light emitted from the lamp is split into two separate beams through the use of a beam splitter. One beam passes through “clean’ air and impinges on the reference photomultiplier tube (PMT) while the other beam passes through the sample and impinges on the measure photomultiplier tube. Data are acquired from the two PMT’s simultaneously. Any drift in the light source is seen on both the reference and measure path. Taking the ratio of the signal strength as observed on reference and measure paths eliminates the effects of light source drift. This methodology allows for common-mode rejection of both short-term noise and long-term drift in the light sources, resulting in exceptional baseline stability. It is this stability which prompted Henderson to write [2] “Accurate measurement of the absorption coefficient at different wavelengths is best made with a double-beam spectrometer”.

Hollow cathode lamps were chosen as the light sources for the analyzers for a number of reasons. They emit characteristic emission lines that are determined solely by the materials used for construction of the cathode. The center wavelengths of the emission lines are known to high accuracy and precision, and due to the nature of the discharge in these lamps, the center wavelength does not drift due to temperature, pressure or any other external influence. The emission band is a fundamental property of the material. For example, a zinc emission line which is used for the measurement of sulfur dioxide occurs at precisely 213.86 nanometers. The bands are also extremely narrow, which contributes to the linearity of the absorbance measurement. These narrow precise and reproducible emission lines are instrumental in achieving high resolution, wavelength accuracy and wavelength precision in the spectrometer, independent of variations in the filter bandpass.

For additional information on the design of UV spectrometers, and factors that influence their performance, the reader is directed to papers presented at previous ISA Analysis Division Meetings [3,4].

Data Processing

The analyzer uses two fixed filters in front of two emission sources. Both lamps are on at the same time, and there is no mechanical chopper in the system. The light intensity from each lamp is modulated by varying the current to the lamp, and thereby varying the light intensity. When one lamp is being pulsed, the other lamp is being “simmered” at a low base current. This simmering effect keeps the lamp warm and prevents transients which can occur when the lamp is first turned on. At the same time, the low simmering current preserves the lamp life.

At first glance, this methodology would appear to be a source of non-linear response, in that two different wavelengths of light are incident on the detector at the same time. However, the unique way in which the data is processed prevents this from occurring. Three separate measurements are taken, one when lamp 1 is pulsed and lamp 2 is being simmered, one when lamp 2 is pulsed and lamp 1 is being simmered, and one when both are being simmered.

It can be shown that the measured transmission (as defined as the ratio between the measured intensity on the sample path as to the measured intensity on the reference path) can be given by:

From Equation 1, it can be noted that the actual intensity of the light source cancels out in numerator and denominator. This indicates that the optical configuration and modulation scheme allows for common-mode noise rejection for light source variations. A similar equation applies for the transmission at the second wavelength. Similar equations may be derived to show that effects of variations in detector sensitivity, and optical path transmission can also be reduced by common-mode rejection.

Thus the mechanical design of the spectrometer and the data processing scheme has been optimized to reduce the largest source of noise (light source variations) and many sources of drift (detector drift and optical path variations due to window fouling).

In addition to these obvious benefits, the analyzer employs an adaptive filter to dramatically reduce noise while not impacting speed of response. This adaptive filter is based on a solid statistical model, and is NOT a predictive filter. The adaptive filter is designed such that it cannot overshoot as predictive filters can.

Testing Results

The analyzer is currently in production and typical test results are presented. During production, each analyzer is tested for ambient zero drift, temperature-induced variations, and linearity.

Typical ambient zero drift for a 50 hr period is presented in Figure 2. We observe that the zero drift over this period of time is less than 0.1 ppm peak to peak. Moreover – there appears to a diurnal component to the drift (most likely due to temperature drift in the building). Typical drift is less than 0.1 ppm in 24 hrs and thus the analyzer can readily be used on a 0 to 10 ppm full-scale range.

FIGURE 2 – AMBIENT ZERO DRIFT DATA

Ambient zero drift data is of course important but in the real world the temperature control in an analyzer building is often not as good as it is in a testing environment. For this reason, each analyzer is subjected to a temperature drift test. In this test, the analyzer is placed in an environmentally controlled chamber and the temperature is cycled from 15°C (59°F) to 30°C (86°F) every six hours. This ensures that each analyzer is tested for temperature-induced drift effects. The test is done at zero gas conditions as experience has shown that the effect on zero is more significant than the effect on span.

Typical temperature drift data is shown in Figure 3. The impact of the temperature change is clearly seen in the reproducible changes in zero-point reading which occur over the 40 hr. period of this test. However, the overall drift over a temperature cycle is still relatively small (approximately 0.3 ppm over 15°C temperature swing) and the cycles tested are much larger than what would be expected in a temperature-controlled analyzer shelter.

FIGURE 3 – TEMPERATURE INDUCED ZERO DRIFt

Baseline noise tests are also performed. In this test, the baseline noise is defined as the standard deviation of the analyzer readings at zero conditions and measured over a 5 minute period. The baseline noise test criterion is a standard deviation less than 50 ppb, with typical results being on the order of 20 ppb.

Typical gas run test results are displayed in Figure 4. In this test, the analyzer was first challenged with 10 ppm span gas, spanned, and then tested at 4 points for linearity. Typically the analyzer will read within 50 ppb of the challenge gas concentration, though in this case there is one point out by 80 ppb. It is important to note as well the fast step change response observed with the analyzer. This is obtained with a sample flow rate of approximately 1200 sccm, and very achievable in stack gas applications.

FIGURE 4 – TYPICAL GAS RUN TEST RESULTS

Summary and Conclusions

New emissions regulations in California have reduced the allowed levels of sulfur dioxide in stack gas by as much as 95%. Many of the installed emissions monitoring systems will not be able to reliably measure sulfur dioxide at such low levels. While ultraviolet absorption spectroscopy has long been a mainstay of such measurements, many existing analyzers are incapable of such low ranges as well. To meet this need, a new direct extractive analyzer has been developed and demonstrated to be able to reliably measure sulfur dioxide in the 0 to 10 ppm range.

References

1. Bennet, H. E., “Accurate methods for determining photometric linearity”, Applied Optics, 5:1265, 1966.
2. Henderson, Brian, “Optical Spectrometers”, Handbook of Optics II, edited by M. Bass et al., McGraw Hill, New York, 1995.
3. Adam, H. and Harris P., “The design and practical application of UV process photometers”, L’analyse Industriel, Paris, France, 1996.
4. Adam, H. and Harris P., “The design and practical application of UV process photometers Part II Optical Components and Configuration”, ISA 96, Chicago, Oct. 1996.