11.5 Data Validation, Reporting & Laboratory Certification

11.5 Data Validation, Reporting & Laboratory Certification

Section 11.4 covered microbiological testing and bench QA/QC. This section covers what happens to the number afterward — whether it is defensible, how it is qualified, and how it reaches SC DES.

---

1. Approved Methods and Laboratory Certification

Compliance samples must be analyzed by a certified laboratory using an approved method for that specific parameter and matrix. A laboratory certified for drinking water coliform is not automatically certified for wastewater BOD — certification is parameter-specific and matrix-specific.

Maintaining certification

• Successful analysis of proficiency testing (PT) samples — blind, known-value samples from an accredited provider, typically twice per year for each certified parameter.
• Periodic on-site audits of facilities, equipment, records, and analyst competency.
• A documented quality manual and standard operating procedures.
• Analyst training and demonstration of capability records.

Failing a PT sample generally requires investigation, corrective action, and a successful make-up PT, and repeated failures can suspend certification for that parameter — which in turn means the plant cannot report compliance data.

---

2. Accuracy and Precision — Not the Same Thing

The classic illustration: a tight cluster of results far from the target is precise but not accurate (a bias problem, often a calibration error). Results scattered widely around the target are accurate on average but not precise (a technique or instrument stability problem). You can have either without the other, and you need both.

Relative percent difference for duplicates

$\text{RPD} = \frac{|R_1 - R_2|}{(R_1 + R_2)/2} \times 100\%$

Worked example. Duplicate TSS results of 46 and 52 mg/L:

$\text{RPD} = \frac{6}{49} \times 100 = \textbf{12.2\%}$

Most laboratories set a duplicate acceptance criterion in the range of 10–20% RPD depending on parameter and concentration.

Percent recovery for spikes

$\%\text{Recovery} = \frac{\text{Spiked Result} - \text{Unspiked Result}}{\text{Amount Spiked}} \times 100\%$

Worked example. An unspiked sample reads 4.0 mg/L; after spiking with 10.0 mg/L it reads 13.2 mg/L.

$\frac{13.2 - 4.0}{10.0} \times 100 = \textbf{92\%}$

Typical acceptance windows run 80–120%. Recovery outside the window signals matrix interference or an analytical problem, and the associated results must be qualified or repeated.

---

3. Detection and Reporting Limits

Reporting rule: never report a result as zero. Report "< MDL" or "ND" with the detection limit stated. Reporting zero implies an absolute measurement that no method can support, and permits often require the detection limit to be shown so that a non-detect can be evaluated against the limit.

Where a permit limit falls below the achievable MDL, compliance is typically demonstrated by reporting non-detect at an MDL at or below a specified minimum level.

---

4. Control Charts

A control chart plots a QC result — spike recovery, duplicate RPD, standard value, or blank — over time against statistically derived limits:

Trends matter as much as single points. Watch for: seven consecutive points on one side of the central line, seven consecutive points trending in one direction, or a sudden shift in the mean. Any of these signals a developing systematic problem — a degrading standard, a drifting instrument, a new reagent lot — before it produces an out-of-control result.

---

5. Data Review Before Reporting

A disciplined operator reviews every result before it leaves the plant:

1. Was the holding time met? (Section 11.1)
2. Was the method approved and the lab certified for that parameter?
3. Did the blanks come back clean? A contaminated blank invalidates the batch.
4. Were duplicates and spikes within acceptance criteria?
5. Is the result physically plausible? Effluent BOD cannot exceed influent BOD. TSS cannot exceed total solids. Dissolved cannot exceed total. Filtered cannot exceed unfiltered.
6. Does it fit the historical pattern? An order-of-magnitude jump is a data question before it is a process question.
7. Do the mass balances close? Solids in should approximate solids out plus solids wasted plus solids accumulated.
8. Are significant figures appropriate? Do not report four digits from a method that supports two.

When a result fails review, the correct action is to investigate and document, then either qualify the result or resample. Never discard an inconvenient result — selective omission of data is falsification.

---

6. Reporting to the Regulator

Signature and certification

DMRs and other regulatory submissions must be signed by a principal executive officer, ranking elected official, or a duly authorized representative — and the operator responsible for the facility is often that representative. The certification statement carries real weight:

the signer certifies under penalty of law, having made inquiry of the persons who managed the system, that the information is true, accurate, and complete, and acknowledges that there are significant penalties for submitting false information, including the possibility of fine and imprisonment.

This links directly back to Section 1.1: falsifying bench sheets, logbooks, or DMRs is grounds for disciplinary action and license revocation under Chapter 51, and carries independent criminal exposure under state and federal environmental law. Report the exceedance. A reported violation is a compliance problem; a concealed one is a crime.