True reflection of the water
Angel Luke of ISCO reports on the selection of automatic samplers
The goal of water sampling is to collect samples that accurately represent the body of water being sampled. Hence automatic sampling devices must be able to:
- collect and store fully representative aliquots of the source being sampled,
- collect a sufficient volume to represent the source, but a small enough sample to be handled effectively in the laboratory,
- collect samples in a way that properly reflects the concentrations of pertinent constituents in the total discharge, l handle samples after collection in a way that does not significantly alter the samples before analysis.
If the samples do not reflect actual flow stream conditions, laboratory procedures cannot compensate to obtain accurate data. This in turn may translate into extra charges or incorrectly reported failure of consent.
Automatic sampling machines differ in their ability to collect representative samples. These variations are due to many factors, including:
- sample intake design,
- intake position in the flow stream,
- velocity of the flow stream.
- sample transport velocity,
- sample transport method,
- vertical distance to travel or lift,
- constituents in the flow stream,
- materials in contact with the sample.
One of the most important factors in accurate sample collection is the transport velocity. Many studies have been made both in the UK and USA to evaluate this criteria and its importance in different sampling techniques.
The official publication on sampling in the USA1 advises selection of a sample pump "capable of lifting a sample a vertical distance of 6.1m and maintaining a line velocity of 0.6-3.0m/s". A publication of the US Environmental Protection Agency2 (USEPA) explains that "sample train velocities should exceed 0.6m/s". Both statements are based on studies made by USEPA’s Harris and Keffer3, who suggest several features of an ideal sampler, including that the sampler should have an intake hose velocity adjustable from 0.61-3m/s with a dial setting.
The suggestion, in turn, is based on earlier recommendations by Shelley and Kirkpatrick4 that "minimum [suction line transport] velocities of 0.6-0.9m/s would appear warranted." It is based upon a formula that shows "a velocity of 0.6m/s is required to adequately transport a 0.09mm particle with a specific gravity of 2.65 and a friction factor of 0.025". This particle is sand. In a 6m suction line suspended vertically, this particle would fall 0.61m during transport. If the suction line used is nominally three-eigths of an inch in diameter, this would mean the first 43.4ml of the sample collected would be devoid of particles with these characteristics. If a 1,000ml sample is being collected, this would result in an understatement of the sand concentration by 4.3%. If the transport velocity were raised to 0.9m/s the error would be 2.9%. At 0.3m/s the error would be 8.7%.
These calculations indicate that achieving accurate samples of a flow stream containing particles with these concentrations requires a transport velocity of 0.6m/s and preferably 0.9m/s. However, professionals who have studied samplers agree it is very difficult — even under experimental conditions — to duplicate results to within ±15%. This means the errors introduced by variations in transport velocity are not significant. In general, only researchers studying sedimentation in rivers and streams are concerned about solids with these characteristics.
The most important application for automatic wastewater samplers is collecting representative samples for analysis of loading effects. Typically these are analysed for BOD, COD, TSS and other factors. The type of solids that contribute to these factors are almost always organic in nature and therefore have densities nearly equal to that of water. Table one lists densities for some typical wastewater pollutants. In this case there would be less settling of solids during transport to the storage containers. The transport velocity would not be a significant factor in obtaining a representative sample. In addition, transport velocity is not a factor when dissolved materials are concerned. In actual applications it has been found that differences in sampling technique and technology do produce varying results.
The differences in the sampler’s ability to take a representative sample was most clearly pointed out in Harris and Keffer’s study. They noticed significant differences in the results obtained from different types of samplers used to sample the influent at WwTWs. Their experiment used two automatic samplers, a vacuum pump and a lower speed peristaltic pump. These units were installed to collect samples from the same point in the flow stream. The intakes were tied together and placed at the midpoint of the influent stream. The collected samples were analysed for BOD5, COD and non-filtrable solids (NFS). Results are listed in Table two, with averages shown in Figure one.
The differences in the sample qualities are clearly due to the different techniques used to automatically collect these samples, the peristaltic system more closely reflecting the values obtained by manual sampling. Some areas that may account for differences in the sample values could be:Cross contamination of samples
The peristaltic pump samples are put directly into the sample container. The portion of the suction line can be purged and rinsed before each sample to ensure the sample collected is free from cross contamination from previous samples. With a vacuum pump sampler, all samples collected must pass and then be retained in a metering chamber. This metering chamber is unable to be adequately rinsed and cleaned between samples. Intake velocities
The intake velocity of the peristaltic pump sampler is closer to the intake velocity of the actual flow stream. This helps to approach the concept of ‘iso-kinetic sampling’ that has been suggested by some researchers. This is where the intake velocity of the sampler and the flow stream velocity are equal. It is theorised this will produce a more representative sample because the flow will not be forced into the intake line. A vacuum pump sampler typically has a line velocity well above the flow stream velocity at lower head heights. This can cause a ‘scouring’ of the channel floor and dislodging solids that have been settled out when the sampler performs a prepurge of the suction line. When the sampler then reverses the sample is collected from a ‘cloud’ of settled-out material that artificially enriches the sample for suspended solids content and renders the sample unrepresentative. Operational differences in the samplers
The peristaltic pump sampler places samples directly into the sample container and meters the volume by means of non-contact liquid sensors. The operator simply selects the volume for each sample to be collected and this volume is then metered through a direct path to the sample container. The vacuum pump sampler must collect a fixed volume of sample, dependent on the metering chamber size. Volumes are then adjusted with the use of a mechanical device inside the chamber to deliver a fixed volume to the sample container. After the fixed volume is inside the metering chamber, the liquid is held in the chamber. This allow solids to settle to the lower part of the chamber. The sample then passed to the sample container may then contain enriched solids due to this settling action.
In all cases of the data shown in the charts, the NFS data was always higher than present in the manual sample. In one case it rendered a number more than 100% higher than the manual sample.
The samples collected by the peristaltic pump sampler more clearly and consistently reflects the results observed in the manually collected samples. The peristaltic pump samples data are shown to be within 5-6% of the manually collected samples data. Variations in the vacuum pump samples can vary as much as 50% of the manually collected samples. This test shows the consistency and accuracy of samples collected using the peristaltic pump sampling technique. Current peristaltic pump samplers are using new techniques and technology to continue to improve the accuracy and repeatability, so that samples are truly representative of the flow stream. This results in:
- better data,
- better informed decisions about treatment efficacy and plant requirements,
- lower effluent charges,
- less possibility of erroneous indication of compliance failure.