Presentation is loading. Please wait.

Presentation is loading. Please wait.

Why do we lose analyzer data? Monitor malfunction DAS malfunction Power outages Environmental problems Wildlife damage Vandalism Operator error.

Similar presentations


Presentation on theme: "Why do we lose analyzer data? Monitor malfunction DAS malfunction Power outages Environmental problems Wildlife damage Vandalism Operator error."— Presentation transcript:

1 Why do we lose analyzer data? Monitor malfunction DAS malfunction Power outages Environmental problems Wildlife damage Vandalism Operator error

2 Why do we lose samples? Sampler malfunction Power outages Storms Wildlife damage Vandalism Site access restrictions Operator error

3 The majority of data loss – be it continuous data or field samples – is due to reasons beyond the control of the site operator!

4 Invalidating Data The site operator is the one most intimately involved with the operation of the monitors & samplers. The QA Section relies on the site operator to do the majority of the data invalidation! The site operator indicates what data needs to be voided on various reports and charts. QA does the actual coding of the data based on the site operator’s documentation. QA voids data that gets past the site operator or if it is a judgment call beyond the purview of the site operator.

5 The QA Section makes decisions on data validity based on the criteria established in CFR and EPA Guidance. PM 2.5 Critical Criteria Table

6 Example of monthly preliminary report submitted by site operator. The site operator invalidated 3 hours of data due to biweekly precision checks and an audit.

7 303 hours of data were actually invalid due to shelter temperature exceedances (highlighted blue). Data recovery = 58% Example of the same data report after it was marked by QA.

8 Example of monthly prelim documented for analyzer drift (negative readings). The site operator also documented troubleshooting and recalibrations with Technical Support assistance. Initial Data Recovery = 92.6%

9 However, 129 additional hours of data were lost due to analyzer malfunction & poor precision results. The negative readings were corrected during the QA process. Final Data Recovery =79%

10 The site operator flagged data as suspicious, but asked QA to make the final judgment call. For this example, the site operator had to replace the analyzer twice during the month, as well as the zero air supply. Initial Data Recovery = 89.5%

11 The site operator did a great job troubleshooting, conducting maintenance, and calibrating during the month. The site operator was correct to flag the data. Because of the equipment malfunctions, there were no precision checks or nightly zero/spans during the time period the site operator flagged as suspicious. Validating data was difficult. Final Data Recovery = 64%

12 Site operator documented lost data due to operator error. Prelim indicates 88.4% data recovery.

13 The same report after being audited by QA. Data Recovery = 94% The “lost” data was recovered during the QA process.

14 So when would the flagging or voiding of data be due to“operator error”?

15 Calibration (SO 2 or O 3 )

16 The site operator documented this chart as a calibration.

17 The results of this biweekly: Precision = 17.5% d. This failed biweekly occurred on July 21. 21-Jul-07 Precision Limits = ±10% d However, the first failed biweekly for this site was actually on June 9, but the site operator did not recalibrate.

18 The recalibration of this instrument finally occurred on August 1. As the chart illustrates, the site operator had difficulty conducting the calibration. All data between June 9 and July 20 was flagged as suspect. All data between July 21 and August 1 was void.

19 ~0.080 PPM 0.025 PPM The chart above shows a substantial drop in concentration when comparing data before the site operator’s visit and after his visit. The site operator did not retighten the mace filter holder after changing the filter inside the analyzer. Five days of ozone data were lost.

20 Pump malfunction on NOx monitor. Failed nightly zero/span at 0200 hours. The site operator did not address this issue for 7 days, despite having received a new pump immediately following the malfunction. The necessary recalibration occurred two days later. A total of 10 days of data were lost. Pump Failure Span

21 The scheduled run date for this sample was August 25. AQS requires concentrations for all scheduled samples. Thus, data for the August 25 sample had to be entered. It was entered as a void. The August 26 sample is considered a make-up. There was no explanation given as to why the filter ran on the wrong date.


Download ppt "Why do we lose analyzer data? Monitor malfunction DAS malfunction Power outages Environmental problems Wildlife damage Vandalism Operator error."

Similar presentations


Ads by Google