TREND ANALYSIS

Many of the models for autocorrelation assume a stationary mean. The expected value, excluding random effects, is assumed to be constant everywhere. That assumption will frequently fail, particularly in the case where we are using yield monitor data to establish a fertility map, assuming that some regions of a field will be generally more or less fertile than other regions. If the mean is not constant, and instead it changes with position, then we no longer can assume stationarity. However, we can account for non-stationarity and removing the moving mean. This may involving fitting data to a trend surface, so we can refer to this as detrending.


Ex .ecutive Smmnarv
The data contained in this report is comprised of Nonconformance Report (NCR), Surveillance Reports, and Corrective Action Reports (CAR). The data population for this report consisted of 277 data records from tile following breakdown:

Nonconformance
Reports 227 Surveillances 45 Corrective Action Reports 5 The data trended now includes six quarters and provides the capability oi' looking at changing patterns in the various performance categories over time. Data pertaining to this quarter is displayed in table form for each subject to summarize current performance.
In previous reports, area charts have been utilized to evaluate each element's contribution to the combined totals. Beginning with this report, the area charts are replaced with line charts.
These charts are shown in two formats, actual and normalized data. The actual chart references each individual contribution without considering variation in the size of the analyzed population. The normalized chart shows the same individual contribution, however variation in population size is factored into the analysis. This allows the standardization of charts and allows the addition of quantifiable goals/limits to be established for each category.
As in the past the basis for determining the elements used in these charts was Pareto analysis of each quarter's data. The data from this individual analysis was then combined and a Pareto analysis was performed to determine major categories for the six reporting periods.
Positive indications include: • A continued reduction in the overall number of deficiencies.
Deficiencies peaked during the Third Quarter 1990 when 800 data records were added. This is the third consecutive quarter for this positive downward trend in total deficiencies.

•
A continued downward trend in causal factors reported as "Indeterminate".
This positive trend is due to an increased effort by the field to determine an initial probable cause for the deficiency when reported. This eliminates the need to make a subjective call as to causal factors durir, g the trend analysis. This category has dropped significantly in overall contribution.
However, it continues to be the major contributing causal factor for Surveillance data. While this is noted as an overall positive trend, more investigation is needed with the Surveillance data and this concern will be forwarded to the Surveillance group for further investigation.
One adverse trend requiring corrective action is evident from this analysis. During the analysis of surveillance data a rise is noted in the finding related to "Procedure Violations". This is the start of an adverse trend and the need to re-emphasize compliance with procedures is needed.
As new and revised procedures come on line it is imperative that timely training is provided and follow up is made to assure personnel understand these changes.
This will establish a solid basis for procedural compliance and avert this condition becoming a significant condition adverse to quality.
Several features will be incorporated into the next Trend Analysis Report,. Future reports will be published monthly, ownership identified at department level and deficiencies detailed to the system and component level. These changes will allow for a more proactive analysis of deficiency data and provide timely feedback for corrective action.

Reactor Res_
Division " Trend Analysis Report Second Quarter 1991

Nonconformance ,Reports
The 227 Nonconformance Reports used in this analysis were distributed among the operating areas as follows: K No significant changes occurred, the same problems in roughly the same order as the five previous reports are noted this quarter. "Installation/Fabrication" accounts for forty four percent of the overall deficiencies.
The majority of these deficiencies are loose or missing nuts/bolts, mis-landed or mis-connected wires. Also note in Chart 2 the consistent decline in the number of reports grouped in "All Others".
This decline is directly attributable to field personnel aiding in the development of clearer more consise problem description categories. : .

Reactor Restart Division "
Trend Analysis Report Second Quarter 1991 It has been noted in two previous reports that "SPAC LTA" continues to decline in its overall contribution to the analysis. This was noted earlier in the report that this category is dropping from significance in relation to NCRs and will be discussed further in the analysis of Surveillances.
There is also a significant reduction in the number of reports that have been categorized as "Indeterminate". This reduction is due to field personnel identit_ng probable cause for the individual reports. This effort has greatly improved the ability to more easily analyze the individual reports.
The rise in "Hardware Failures" is directly attributable to the Startup Testing program currently in progress in K Area.

• Reactor Restart Division -Trend Analysis Report Second Quarter 1991
The dispositions prescribed for the NCR deficiencies are: A marked improvement is noted with the reduction in the "None Notcd" category.
This is a significant step in identifying corrective actions which have a positive impact on recurrence prevention. "Modify Design" shows a rise in occurrences due to the number of drawings being revised to show "as built" configurations. The "All Others" category has been dropped from this report. This is due to several new categories being added at the request of Site QA. As a baseline is established using these new categories these charts will be updated in future reports to reflect actual performance. The rise in the number of "Procedure Violations" noted in this report is the start of a negative upward trend.
The importance of procedural compliance needs to be re-emphasized.
A direct correlation with this finding category will be shown in the following section when causal factors will be discussed. Tile "All Others" category has been dropped from this report. This is due to several new categories being added at the request of Site QA. As a baseline is established using these new categories these charts will be updated in future reports to reflect actual performance. Thirty-one of forty-five identified causal factors identified correlate to "Procedure Violations". _his is 69 percent of the data population used in this report's analysis. This information will be directed for management attention and will be analyzed further. This is the beginning of a significant adverse trend.  The corrective actions identified from the surveillance reports shows steps ar_ being taken to eliminate recurrence of identified deficiencies.
As new and revised procedures come on line it is imperative that timely training is provided and follow up is made to assure personnel understand these procedures.
This will establish a solid basis for procedural compliance and avert this condition becoming a significant condition adverse to quality. In all cases a lack of adequate procedure or administrative policy is cited as the main causal factor. Steps are being taken to develop or revise the control systems needed to prevent recurrence of these conditions.