Top Ten BOD/CBOD Glitches
by Perry Brake
It’s hard to narrow the possible glitches in the BOD/CBOD test to a “top ten”, but I would like to give it try. They are listed below more or less in the order I would check them if reviewing how well a lab is able to do the test.
- Blanks. If dilution water blanks are not under control (the method says (0.2 mg/L, but the goal should be 0.0 mg/L, with an occasional 0.1, and a rare 0.2 mg/L), whatever is causing the blank problem can negatively affect both bias [1] and precision [2], the two components of accuracy. If total precision is good (indicated by a standard deviation of GGA results of 15 mg/L or less), bad blanks are almost always caused by choice of source water (e.g., distilled, deionized water). If total precision is not so good, bad blanks can be caused by any of many factors such a problem in measuring dissolved oxygen (DO) levels, cleanliness of labware, and photosynthesis during incubation.
- Precision. Once blanks are under control, the next thing to check in troubleshooting the BOD/CBOD test is precision (see definition in Footnote #1). Many factors, such as bad blanks, variability of seed or GGA solution quality, malfunctioning DO meter, and sloppy lab work, can influence precision. Standard Methods indicates the standard deviation of repeated GGA analyses should be 10 mg/L or less. While that level of precision is achievable, such precise work is probably not necessary to obtain reliable results. I recommend 15 mg/L or less, the same value EPA suggests for CBOD (but the concept also applies to BOD) when they say the percent relative standard deviation (standard deviation divided by the average and expressed as a percent) should be 7.5% or less. For example, if a lab’s GGA average is 198 mg/L, 7.5% of that would be approximately 15 mg/L.
- Bias. Only after blanks and precision are under control should an attempt be made to solve bias problems. If precision is good, excessive bias (i.e., an average far above or below the desired level of 198 mg/L for BOD) is almost always caused by the choice seed (see #4 below).
- Choice of Seed. Choice of seed can have a dramatic influence on both bias and precision. For example, lyophilized (freeze dried) commercial seeds usually give precise (i.e., consistent), but low results (the manufacturer’s preparation instructions can often be tweaked to increase results). A wastewater treatment plant influent usually give a high average GGA, but results are often imprecise (inconsistent) because of influent variability caused by season, local weather, municipality activities, and other reasons. Effluent from the primary process at a domestic wastewater treatment plant usually yields an ideal seed, being consistent and having a high population of suitable bacteria. Plant effluent before disinfection is consistent, but usually weak (effluent after disinfection is not acceptable).
- Source Water. Quality of source water may cause problems with blanks and/or GGA results. Ammonia and some organic materials can go through a still resulting in high blanks. Organic materials can be leached from deionization columns…again, causing high blanks. In either case, if GGA results are near being unacceptable on the high side, the organics and/or ammonia could push them over the brink. Dissolved metals can cause low blanks and cause the average GGA to dip below a lower threshold. Sometimes, bottled distilled water will provide good blanks…”steam distilled” distilled water is probably the best reasonably available bottled water, while distilled drinking water is probably the worst because of materials added for “taste”.
- Measuring DO. Accurate measurement of DO is essential to accurate (unbiased, precise) BOD/CBOD results. Winkler titrations, while very accurate if done properly, are time consuming and generally not necessary to achieve good results. If using a DO meter with membrane, it is important to use actual atmospheric pressure (and not the elevation of the lab) and temperature when calibrating. Bottle contents (including the calibration bottle) must be at 20 ± 1° C for best results. Each bottle must be just saturated with DO at the beginning of the incubation (neither supersaturated, nor under saturated).
- GGA Preparation/Purchase. Labs having a problem with achieving unbiased and precision GGA results are sometimes quick to blame the GGA solution itself. In reality, the solution is seldom a problem if prepared according the instructions in the method or purchased from a reputable provider. If there are GGA problems and the lab is preparing its own GGA solutions, it would be worthwhile to try any of several commercially prepared standard. It is important to use a GGA solution that is the proper concentration (150 mg/L glucose, 150 mg/L glutamic acid) to assure legal defensibility of data.
- Number of Bottles per Sample. Many labs analyze two or more bottles for every sample (e.g., blanks, seed control, GGA, effluent, influent and others), often “because it has always been done that way” or because of the mistaken belief that the average of two (or more) results is “more accurate” than any single result. If total precision is good, there is no valid scientific reason for running more than one blank, seed control, GGA, or (usually) effluent because their properties vary little if any from day to day. Influent samples must be analyzed in more than one dilution to make sure at least one bottle meets the “deplete at least 2.0 mg/L, retain and least 1.0 mg/L) method criteria. Time saved by doing fewer bottles can be used paying more attention to #10 below. The publisher of Standard Methods recently said in writing that Method 5210B never meant that multiple bottles must be analyzed for a well-characterized sample.
- Interference. Some materials (e.g., septage) interfere with the ability of seed bacteria to do their job. This can be detected by doing several dilutions (I recommend at least seven), the most dilute of which include a 10-fold (or more) dilution of the original sample. If BOD/CBOD values goes up as the sample becomes more dilute, something in the sample is interfering.
- Monitoring Performance. BOD/CBOD testing can go awry in many different ways making it important that labs monitor performance of the test closely. Use of a control chart is the simplest and most effective way of doing that. The average (middle) line on the chart indicates bias in the test. For example a 210 mg/L average for BOD would indicate a high, but acceptable, bias…an average of 150 mg/L would be unacceptable. The standard deviation used to construct the chart indicates starting precision…15 mg/L or less is acceptable. The spread of results subsequently plotted on the chart indicate ongoing precision. An occasional duplicate can be used to check within-batch precision, but to do duplicates with every batch is probably not worthwhile unless the lab is having a precision problem and wants to see if within-batch imprecision is a significant factor. So there you have it…a “Top Ten”. Could I go on? Sure, but these will probably get you pointed in the right direction. Perry BrakeAuthor of A Bug’s-Eye-View of the BOD Test”, available from Environmental Express.
[1] For the purposes of this blog posting, bias is an indication of systematic error, and is estimated by comparing the average result for repeated analyses of the glucose/glutamic acid (GGA) standard to the accepted value of 198 mg/L for BOD.
[2] For the purposes of this blog posting, precision, or technically imprecison, is caused by random error resulting in spread of results above or below the average result. Total imprecision is estimated by calculating the standard deviation of repeated results. Within-batch imprecision, estimated by doing duplicates, and between-batch imprecision determine total imprecision (between-batch imprecision cannot be measured directly.
