Exhaust Gas Sampling and Conditioning

Heinz Burtscher, W. Addy Majewski

This is a preview of the paper, limited to some initial content. Full access requires DieselNet subscription.
Please log in to view the complete version of this paper.

Abstract: Before emissions can be measured, gas has to be sampled from the exhaust system. In most cases, gas is sampled from diluted exhaust. Exhaust gas dilution can be performed either in a full flow dilution tunnel, such as the CVS system that is commonly used for regulatory testing, or in a partial flow dilution system. The dilution step is very important to obtain representative measurement; sample losses or changes in sample properties can cause significant errors.


Before the exhaust gas and its components can be measured, it has to be sampled from the exhaust system. The sampling system acts as an important interface between the exhaust and measurement systems. Its role is to pre-condition the sample to meet the requirements of the particular measurement instruments—such as temperature, concentration, or the presence of moisture and volatiles—while preventing or at least minimizing physical and/or chemical changes of the measured species. Most measurement instruments are designed to operate close to ambient temperature. This means that the hot exhaust gas has to be cooled to an adequate level.

Diluted Sampling. As the exhaust contains volatile material (water, sulfur compounds, VOCs, ...), cooling may lead to supersaturation of many of these species, resulting in condensation and nucleation. To avoid this, the exhaust gas has to be either (1) diluted to a level where supersaturation can be avoided or (2) the volatile material has to be removed from the exhaust. Dilution—which is also an effective means to obtain the required temperature reduction—is the preferred method in most exhaust gas measurements. Dilution may be also required to lower the concentrations to match the ranges of the instruments used.

The sampling requirements for particle and gas species measurement are somewhat different. For gas sampling it is of crucial importance that no chemical reactions take place in the sampling train. This means that inert materials have to be used for all parts coming in contact with the exhaust gas. This is often achieved by using teflon and stainless steel tubing. On the other hand, nucleation, condensation and residence time in the dilution tunnel are usually less important in gas measurements. If condensation occurs, water traps can be used prior to measuring. This also means that the influence of temperature is less pronounced.

An important function of the gas sampling system is to condition the sample to meet the inlet sample specifications of the particular analytical instruments. Gas analyzers typically require that particulate matter is removed from the sample. Thus, particulate filters are necessary components of the sampling system. Some of the analytical instruments (so called dry instruments, such as most infrared (IR) analyzers) require that the sample is cooled and dried. Water vapor should be condensed and removed from the sample in a way that minimizes losses of acidic gases (NOx, SOx) due to their solubility in the condensate. This can be achieved through very rapid cooling in Peltier coolers. Several instruments—including FID hydrocarbon analyzers and chemiluminescence NOx analyzer—are also available in heated (wet) versions to prevent losses due to condensation in the sampling system. All sampling system elements which are in contact with the sample (sampling lines, pumps, filters, valves) are also heated and maintained at temperatures of about 180-200°C.

In particulate matter sampling, nucleation and condensation may significantly change the particles. The sampling conditions therefore have to be well defined to either avoid nucleation/condensation or at least keep it under control. This is usually done by keeping the temperature in the dilution tunnel within certain limits. Coagulation presents another issue in particle sampling. Particles are subject to collisions due to thermal motion. Unlike gas molecules, particles stick together with a probability close to unity after collision, which causes the particle size distribution to continuously change towards larger particles. Therefore, to obtain representative results, the time the particles remain in the sampling system has to be small enough to keep this effect negligible. The coagulation rate scales with the square of the number concentration. Immediate strong dilution therefore is a powerful way to reduce the influence of coagulation. These topics will be discussed in more detail in the following sections.

Raw Gas Sampling. In spite of what has been said above, undiluted or raw gas sampling is also a common technique because of the relative simplicity of the sampling system. However, because of the high concentration of moisture, particulates, and high temperature, it is not an easy task to obtain reproducible and reliable results. While allowed for some regulatory applications, it is mainly used for non-regulatory and/or field testing.

Instruments such as smoke meters or opacity meters can handle the high concentrations of undiluted exhaust constituents and can be operated at elevated temperatures to avoid condensation. These instruments can therefore be used with an undiluted sample. Opacity meters, however, run into problems with the detection limit for modern low emission engines.

For a number of nonroad applications—for example stationary power generation plants or marine diesel engines—undiluted sampling is used to collect filter samples to determine particulate mass. The filter samples are then conditioned to account for the effect of the volatiles. The US EPA also allowed undiluted sampling for the measurement of gaseous emissions from mobile nonroad engines such as those used in construction equipment (Tier 1-3 standards). However, PM samples still had to be taken from diluted exhaust.

Isokinetic Sampling. When taking a sample from a gas flow (diluted or raw), ideally the flow velocity in the sampling line should be the same as in the flow, a condition referred to as isokinetic sampling. Otherwise, aerodynamic effects may lead to a depletion or an enrichment of coarse particles in the sampling line [645]. The importance of this effect increases with increasing particle size. For the size range of interest in diesel emissions, errors due to non-isokinetic sampling are usually negligible.