2. 1 - 3.6. FIELD AND LABORATORY WORK 3.6.3. Quality Assurance and Control (QA/QC)

3. Peter Kelderman UNESCO-IHE Institute for Water Education Online Module Water Quality Assessment 2

4. 3 Quality Control should take place in all steps of monitoring cycle Defining objectives monitoring; optimum network, frequency,.... We discussed above items e.g. in Course 3.4. (optimization) Here we will mainly discuss QA/QC in a laboratory

5. 4 In certified laboratories some 20% of resources spent on QA/QC Well-trained, motivated staff, good facilities (electricity, Airco, cleaning, etc.) Daily Standard Operating Procedures (SOP) for whole monitoring cycle; use of “Standard Methods”* Good Laboratory Practices (GLP) (cleanliness, accurate working, no smoking, etc.) Good reporting of data Frequent control of staff, methods, instruments. *Much used: APHA-AWWA-WEF: Standard Methods for the Examination of water and wastewater (new editions every 3-5 years)

6. 5 Accuracy and precision of data set “Outliers”: see later Rounding off: present the results according to the accuracy/precision of the method. E.g a pH of 7.32 should be rounded off to 7.3 Concentrations below detection limit (D.L): report as zero or <.. (see Standard Methods). E.g. the D.L. for PO 4 -P will be around 0.005-0.01 mg P/L (5-10 μg P/L). SOME CONCEPTS

7. 6 Accuracy and Precision of Data set Scattered data, incorrect average Scattered data; correct average. Random errors due to careless working. Non-scattered data; incorrect average. May be due to systematic errors (e.g. instrumental) Non-scattered data; correct average

8. 7 Internal control (daily): -Blanks, duplicate samples, “spiked” samples. Care should be taken that the Control samples are not recognizable by the lab staff. -- Check of ionic balance, Cl - /Na + ratio, PO 4 -P <P tot., etc. - Use of “Control Charts” Keep the samples few weeks for possible re-analysis; also good reporting by the sampling crew  Field notebook. - QA/QC

9. QA/QC checklist (1) Does laboratory personnel have: — clearly defined responsibilities? — qualifications, experience; training? Is lab space: — adequate for the types and number of analyses being undertaken? 8 Is equipment: — adequate? — regularly serviced and maintained? — calibrated and used only by authorized personnel?

10. 9 Are materials: — bought from a reliable supplier, who carries out quality control? Are there proper facilities: — for the receipt and storage of samples, and systems for coding and identifying them? QA/QC checklist (2) Are data: — archived? — retrievable?

11. QA checklist (3) Are methods: — validated? — documented? — monitored (i.e. the results subjected to analytical quality control)? 10 Is safety assured by: — adequate working and waste-disposal procedures? — training of staff? — proper maintenance of equipment? — proper supervision of staff?

12. 11 FIELD NOTEBOOK Data on site, date/time, weather conditions,.. Sampling data; preservation, transport,.. Analyses on site Persons involved  later checking possible !

13. 12 Shewhart Control chart for Zn, constructed using the x avg. and s x * for control samples in the lab. Probability for x avg. + 2 s x : 5%*  “warning limit” Probability for x avg. + 3 s x : 0.3%*  “action limit” * See Course 4.2.

14. Trends Shewhart Chart and interpretation 13 1. Incorrect calibration due to bad standards or malfunctioning of equipment. Or: impurities. 2. New standards of other quality used. 3. Degradation of the calibration standards. 4. Technical error or insufficiently experienced/trained) personnel.

15. 14 Practical criteria for taking actions, e.g.: One value outside action limit or two consecutive values outside warning limit 7 consecutive values increasing or decreasing 10 out of 11 consecutive values on one side of the line Actions: repeat analysis immediately. If correct, find out reason of bias: Systematic/random errors? (Instrumental? Chemicals?; Lab staff? Procedure?...) ACTIONS

16. 15 External control (few times per year): - Use of “Certified reference materials” (CRM) -Inter-laboratory checks (“Ring” or “Round Robin” tests), for performance testing individual labs, or errors between labs. “Standard Methods” often gives precision and accuracy of a method reachable within one lab, or between labs. In general: Critical but open atmosphere lab staff/supervisor Optimum quality (100% reliability not possible)

17. 16 Some “Tools” for finding out (in)consistencies: Ionic balance: Σ + milli-equivalents/L = Σ – milli-equivalents/L. (milli-eq/L = |milli-mole/L x charge| ) Example: A water contains 40 mg/L Ca 2+ /L, 23.0 mg Na + /L and 106.5 Cl - /L at pH = 7.0 (so [H + ] = [OH - ]). Is the ionic balance correct? (Atomic masses Ca=40.1; Na = 23.0; Cl = 35.5) 40.0 mg Ca 2+ /L =40.0/40.1 = 1.0 mmole/L = 1.0 *2 = 2.0 meq/L 23.0 mg Na + /L = 23.0/23.0 = 1 mmole/L = 1.0 meq/L 106.5 mg Cl - /L = 106.5/35.5 = 3.0 mmole/L = 3.0 meq/L  Σ- ions = 3.0 meq/L So the ionic balance is correct (in practice an error of around 5% is acceptable)  Σ+ ions = 3.0 meq/L

18. 17 Some “Tools” for finding out (in)consistencies (cont’d): Check for possible interchanging of data columns Check for outliers (see later) Think of correctly rounding off PO 4 -P <P tot. ; NH 4 + < N tot., etc. Repeat sample analyses for questionable results Use unbiased “expert opinion”

19. 18 EXAMPLE INTERNAL CONTROL (see Chapman, 1996)

20. 19 Sample 1: Ionic balance correct; correctly rounded off, etc.  results okay Sample 2: Excessive number of significant figures in Ca 2+, NO 3 -N, PO 4 -P, pH Sample 3: High Na + and Cl - : outliers? Also not in line with low EC  re-analyze

21. 20 Samples 5 and 6: incorrect ionic balance  re-analyze Sample 7: Results of SO 4 2- and HCO 3 - have been interchanged Sample 4: Mg 2+ probably 10 x too high  correct this

22. 21 Assignment.@@: Water quality data in the Kirinya wetland, Uganda Give your expert opinion on the reliability and consistency of the data: Which data would you leave out or modify? HINT: reject as few data as possible; “errors” must be really clear ! In practice, always keep the original data set as well, for later re- inspection. Please put your expert opinion on the platform

23. 22 nQ crit 30.970 40.829 50.710 60.625 70.568 80.526 90.493 100.466 150.384 200.342 250.317 300.298 Outliers in Kirinya data set?  Use e.g. Dixon Q-test In data set: 1, 3, 2, 3, 2, 4, 3, 5, 2, 11 value “11” may be outlier. Test should be applied, e.g. Dixon Q-test. Distance between suspected value q s “11” and neighbouring value q n “5” :11- 5 = 6. campare with width of data w: w = 11-1 =10 Reject “11” if holds: 6>0.5 x 10; true  reject ! In general: reject if holds: |q s -q n |> Q x w For Q: see Table, as function of n.

24. 23 Other aspects of Quality control See e.g. Book “Standard Methods” for correct procedures, preservation, possible errors, etc. Quality control in whole monitoring cycle; don’t neglect QA/QC in the other steps ! Lab accreditation according to ISO standards For (optional) self study: “google” for these ISO standards for laboratories, e.g. ISO 5667. Additional reading; see MTM Reporting on QA/QC is also necessary for Reports and scientific journals.

25. .......Additionally five representative samples were taken from the groundwater spring. Water samples were collected manually in 100 mL polythene bottles, filtered over a 0.45 μm filter and stored at 4 0 C. Subsequent analyses took place generally within 2 days at the laboratory of the Kenya Bureau of Standards in Nairobi; this laboratory has been accredited according to ISO 17025 Standards by the United Kingdom Accreditation Services (UKAS). Water samples were analysed for Soluble Reactive Phosphorus (SRP) according to the ammonium molybdate method (Eaton et al., 2005; for details, see Koech, 2006). Quality Control on the analyses comprised analyses of a limited number of reference samples, of blanks (one per sample) and duplicate samples (one per 10 samples) (acceptance criteria of the duplicates ± 10 % of the arithmetic mean).... Phosphorus budget in a low-income, peri-urban area: Kibera, Nairobi (Kenya) P. Kelderman, D. K. Koech, B. Gumbo and J. O’Keeffe (2009). From: Water Science and Technology, 60 (10), 2669-76. Example:reporting on Methods; QA/QC in a scientific journal