QED Correlation for TPH

Correlations between different laboratories, even the same laboratories for duplicate samples is often difficult for TPH analysis because of the nature of the contamination. Petrol, JP-4, Kerosene, JP-5, JP-8 and fresh diesel contain a high percentage of volatile components that during sampling, sample transportation and subsequent sample preparation, can be lost. If the lab does not detect these volatile compounds, typically the BTEX, monoaromatic and naphthalene compounds, it does not mean the on site analysis that does detect them is wrong.

The other significant cause of variability is due to sample homogeneity. Due to the volatile nature of many hydrocarbons, it is difficult to mix soil samples to evenly disperse the hydrocarbon without causing the volatiles to be lost. Most laboratory extraction systems can only handle between 5g and 10g of soil, which is a small proportion of the 200g - 500g quantity of sample sent in. The statistical significance of taking a non homogeneous sample and analysing only 5% of the total sample and expecting that to be truly representative of the sample, let alone the 20m x 20m x 0.5m volume of soil that the sample is a representation of, is always going to be tenuous. (see Uncertainty section)

To this conundrum, the nature of TPH analysis should be applied (see TPH analysis section). Different laboratories use different techniques and derive different TPH values depending on the type of hydrocarbon that is present in the sample. The UK MCerts accredited laboratories are allowed a 30% variance from the certified result from the round robin performance samples and still retain their accreditation. This is for a sample that is fully homogenised, contains little volatile material and is of a known hydrocarbon type. Real samples are very likely to have a much larger variation.

The correlation data for QED compared to laboratory shown below is from a former fuel storage site that had been bio-remediated. The samples analysed by QED are taken from the same trial pit as the laboratory analysis samples, but are not true duplicates. The laboratory samples were placed into amber wide mouth screw top jars, filled to the top and put in a cool box for shipment to the laboratory to arrive the next day.

Sample 24 TPH value QED 69 mg/kg Lab<38 mg/kg
This has acceptable correlation for TPH, but the lab result for the sum 16 PAHs shows an anomaly. Fuel hydrocarbons contain very low concentrations of the 16 target PAHs relative to the other compounds present. To find that the concentration of these individual compounds is greater than the total TPH is surprising. The laboratory uses a different method to extract and analyse for the target 16 PAHs. This means the portion of sample analysed for the DRO/GRO and total aromatics is not the same as the sample analysed for the target 16 PAHs and the problem of sample homogeneity can arise. In situations where the lab target 16 PAH result is more than 30% of the lab total aromatics value, poor sample homogeneity should be suspected.

Sample 5 TPH value QED 5,420 mg/kg Lab 13,268 mg/kg
QROS ID = degraded diesel. The amount of GRO relative to the DRO found in the QED sample indicates a relatively low amount of degradation. The lab data indicates the sample is diesel from the ratio of C16 and C20 aromatics/aliphatics, but with no GRO compounds detected, indicating how quickly the GRO range can be lost if samples are taken into simple amber jars. The significant difference between the two results is however most likely due to sample homogeneity. The matrix ranged from soil containing small rock fragments to material predominantly containing rock fragments. The QED analysis only takes samples containing fragments smaller than 5mm where possible and does not compensate for moisture or stone content. The ratio of total aromatic to DRO for both QED and the lab indicate that the hydrocarbon is a similar type further indicating sample homogeneity may be the major cause of the discrepancy between the two values.

Sample 6 TPH value QED 11,742/2,279 mg/kg Lab 5,874 mg/kg

2 separate samples were taken from this trial pit at the same depth. Only one of the samples was sent to the lab. The QED results show the potential variability in the soil homogeneity over a very small area. The QED fingerprints also show the material in the samples was not the same. The average value is close to the lab value however. The sample identified as 6 (1) showed a higher proportion of GRO and lower molecular weight compounds than sample 6 (2) indicating a much less degraded material, but both fingerprints indicate degraded diesel.

Sample 10 TPH value QED 323 mg/kg Lab 1,183 mg/kg
The QED fingerprint shows a predominantly bituminous type material, the general classification of hydrocarbon that is no longer recognised as a fuel and shows very significant degradation with the loss of nearly all aliphatic components. There are indications of plant material such as humic acids as well. The lab data shows the material is not diesel from the very low concentration of aliphatic compounds relative to the aromatic compounds in the sample result. The lab TPH value is derived almost entirely from the aromatic values in the EC21 – EC35 range, which is a good indicator of humic acids and other plant derived material when no corresponding aliphatics are present. Humic acids are not derived from petroleum and should not be included in the TPH result. The lab will calibrate the TPH analysis against diesel and lubricating oil standards which are predominantly aliphatic compounds. The result obtained from a predominantly aromatic substance, especially compounds in the aromatic EC21 – EC35 range may be significantly over-estimated relative to the actual concentration of material in the sample using this calibration method.

TP7 TPH value QED 221 mg/kg Lab <38 mg/kg
The QED fingerprint indicates a degraded fuel with a close match to the degraded fuel calibrator (right). The significant variation in results between QED and the laboratory is possibly due to the sample homogeneity.

Sample 22 TPH value QED 26,056 mg/kg Lab 7,768 mg/kg
The QED fingerprint shows a slightly degraded diesel with GRO present. The dilution required to get the sample within the QED calibration range was very high, indicating a high concentration of material was present in the sample. The laboratory shows traces of BTEX and MTBE in the sample, indicating petrol is present. The low result from the lab compared to QED cannot be attributed completely to losses of volatile compounds, although this is likely to have occurred. Sample homogeneity may again be the most likely cause of the difference in results.

Sample 18 TPH value QED 428 mg/kg Lab 456 mg/kg
The QED fingerprint shows a degraded gasoline. The correlation between the lab and QED is excellent, but there are differences in the GRO, DRO and total aromatic values generated by the two methods. QED is more likely to correctly measure the aromatic compounds because the aromatics found in products derived from crude oil fluoresce in the UV and this fluorescence is measured by QED. The lab method detects anything that comes through the GC column, regardless of wether it is aromatic or aliphatic and the clean up method used to separate the aromatics from the aliphatics is not perfect. The lab results show a significant loss of BTEX, but the GRO reported by QED may be included in the lab DRO.

The results for this project demonstrate the wide variety of hydrocarbon fuel types that can be encountered on a relatively small site and that sample homogeneity can be a significant issue for obtaining correlations, but that overall the laboratory and the QED analysis broadly agree and the QED is effective in identifying the hydrocarbon type. The QED identifications agreed with the site plan as to the location of the various storage tanks and the expected hydrocarbon in the soil.

The correlation was deemed highly acceptable by the client and the differences in results had logical and scientifically valid reasons. The data does however show that higher sampling densities will highlight potential homogeneity issues and help minimise the potential of an inaccurate Conceptual Site Model from being developed. On this project, 196 samples were analysed by QED over a 6 day period using one chemist to analyse the samples and one environmental scientist to map the sample locations and collect the samples. The correlations for the sum of the 16 target PAHs was also acceptable.

The data in the box to the left shows the effects of storing sample in the solvent over a period of time on site with no other preservation, leaving a sample (Test 3) known to contain gasoline in the standard laboratory sampling jar for several days and Test 6 to demonstrate the reproducibility of the method.

The results show that for petroleum hydrocarbons, the extraction is very efficient and no significant increase in hydrocarbon is seen. The extraction solvent also shows that it is a good preservative for volatile compounds, with little loss of the GRO fraction over 4 days.

Test 3 does show how significant the losses of volatile compounds can be on an undisturbed sample in an amber jar with an ambient temperature of not above 12 degrees Centigrade even at the hottest time.