Back to the Basics: Use of references gases in calibration

The practical difficulty in accurately preparing and storing exact solvent concentrations often precludes their use. Multiple solvent mixtures are even more difficult to adequately control. 

Back to the Basics: Solvent Calibration

Unfortunately, the process of preparing accurate solvent mixtures for the precise calibration of the analyzer under the actual operating conditions is sufficiently difficult and unreliable in many cases, to generally prevent calibration with solvent mixtures. 

Back to the Basics: Adjustment of Calibration for Process Temperature

Testing by the authorities has shown that at elevated temperatures the Lower Flammable Limit decreases, that is, the mixture increases in flammability. 

This means that extra care is required in the calibration of analyzers used in heated industrial process above and beyond that typically performed for ambient temperature leak detectors. 

The calibration data obtained for a particular solvent should always include the temperature correction, or give some basis for making proper temperature correction.


Back to the Basics: Effects of Varying Rates of Evaporation on Calibration Accuracy

For multi-solvent, multi-zone drying processes, it is not always proper to assume uniform evaporation of solvents in each zone.

For an analyzer with wide variation in response factors to the solvents of interest, there could be a considerable error from the assumption that the solvent mixture is uniform in each zone of the dryer. Potentially, the relatively volatile solvents could exist in greater proportion in the first zone(s) of the dryer, and the less volatile solvents could exist in greater proportion in the later zone(s) of the dryer. 

Back to the Basics: Calibration Data

In verifying the analyzer's range of response to different solvents, it’s best to obtain calibration data based on the response to solvent concentrations expressed in terms of LFL. 

Do NOT use the following:

  1. Response factors based upon weight-percentages (such as milligram carbon per normal cubic meter), these may be deceptively close in response until they are translated into terms of %LFL
  2. Calibrations which are based upon calculations and have not been empirically tested and proven using actual solvent mixtures

Back to the Basics: Responses to Various Solvents

An analyzer should respond uniformly to all solvents that might be used. Any differences between individual solvent response factors should not reduce the margin of safety. Typically, this means that the calibration is based upon that solvent producing the lowest response, so the analyzer indicates the true concentration of this one solvent and it indicates readings higher than actual concentration for all others in use.

Back to the Basics: Industry-Standard Accuracy

Several analyzer types are of sufficiently sound design and manufacture to meet the industry-standard requirements for general purpose gas detection. Many of these also maintain this accuracy under normal conditions for a reasonable period of time following calibration.

It is much less common, however, for the accuracy to be maintained for the following, that are routinely found in industrial processes:

Back to the Basics: Safety and Economy

In monitoring the solvent concentration in industrial processes, the accuracy of the analyzer directly affects the safety and economy of the process.

There are two types of errors to look out for:

Back to the Basics: Flash Point, Dew Point & Temperature

The ability of the analyzer to measure the solvent concentration depends on the delivery of the solvent vapor to the analyzer through a sampling system in which all elements in contact with sample are above the dew points and flash points of all substances in the process. 

Back to the Basics: Flash Point and Dew Point

Each solvent has a characteristic “flash point.” This is the minimum temperature at which the solvent can produce sufficient vapors to form a flammable concentration in air at its surface. The flash point is determined by several test methods, giving some small variation in the published values.