ML20059C584
| ML20059C584 | |
| Person / Time | |
|---|---|
| Issue date: | 09/16/1993 |
| From: | Fields J, Young B NRC OFFICE OF ADMINISTRATION (ADM), SCIENCE APPLICATIONS INTERNATIONAL CORP. (FORMERLY |
| To: | |
| References | |
| CON-FIN-L-1384, CON-NRC-04-91-054, CON-NRC-4-91-54 NUDOCS 9311010209 | |
| Download: ML20059C584 (11) | |
Text
b
- 1. CONTRACT 80 COOL
"/Agg opg.ggg AMENDMENT OF SOLICITATION / MODIFICATION OF CONTRACT I
I2
( Rcauisn.ONfeuRCnAse Rw. NO.
- e. eROscCi NO. gf o,,,uc.eu; inyc pAi c 2.terg/ / D /@
RFPA RES-91-054 dtd. 8/11/M
,. AucNOuun/uoo.. cAv ON No.
D Five (5)
- r. AOuiret cRcQ uv or omer man e== di cogg l f
j
-w,33acD ov U.S. Nuclear Regulatory Commission RES Document ID #RES-C93-469 Division of Contracts & Property Management Washington, DC 20555 g
eA. AMLNOMLNT OF SOUCH AYSON NO.
A NAME ANO ADORCSS OF CONT RACTOR (Ao, #Wef. cound~5s.k and MPCodef Science Applications International Corp.
Attn: Brenda Young, Contracts Manager eu.oattO p u n u u) 301. Laboratory Road, P. O. Box 2501 Oak Ridge, TN 37831 noa. MOO F CATION Oe cONTaAci/OROc NO.
Principal Investigator / Technical
Contact:
Mr. Derek Haopmover na.DAi m ssu na us Telephone No. (615 )
48P 0011 8/5/91 lFACIUTY CODE CODE 11.THIS (TEM ONLY APPLIESTO AMENDMENTSOF SOLICITATIONS O ue tenid. O is oot e, O 2t..
e numeene sencitation is amended.e. eet forth in item i4.we.-,eed oste spacined f r
- pt of Of,e,.
Ofters must ednowledge receipt of this amerdment prior to the hour and date specfood in the concitatio tended.
f copies c( the amendment: (b) Dy ecknowledging receipt of sMs amendenent on ee& c submitted; or Ict By esperate letter or telegram which includes e reference to the solicitation end amendment n (a) By conpleting items 8 and 15.end retuming _
MENT TO BE RECEtVED ATTHE PLACE DESIGNATED FOR THE RECEIPTOF O i
letter,piJed each 4 or letter snakes soference to the concitation and this a
- 12. ACCOUNTsNG AND APPROPR4ATION oA1 A (J/ required)
$112,674 L1384 36019202400 BOC: 2542 31X0200.360 Amount nhlinatnet.
- 13. THIS ITEM APPLIES ON LY TO MODIFICATIONS OF CONTRACTS /OROERS.
IT MODIFIES THE CONTRfy/ ORDER NO. AS DESCRIBED IN ITEM 14.
" '0 !'T MYR. "" "'"^"" ' "***"***" "' '"^"' "* " * '" " '" 2 ' ^"' "^"' '" * "' ' "'
'm ^WC'EL"#R NO
'* "^" $
t N103 b
^ "
% ^,A".<TM.D ic"fscT?OItm'iI/?T"ch"e'.5UEffA
"^
e T
H TY C. THis SUPPLEMENT AL AGREEMENT 45 ENTERCO GNTO PUl%UANT 10 AUTHOR 4TY OFs X
Changes-Cost Reimbursement, Alt. No.1 (FAR 52.243-2)
O.OTHER (Specify type of modd/ketion end authority) j P
. cogies io ing istuins orrice.
E.IMPORTANT: Contractor O i not. O is cenoi, a io.isn ini oocument.na,1oro 1 OCSCRIPT 80N OF AMLftDMENT/MOOffICAT4ON (Orr.ndred try UCreeccion beedenss,inciuding soucidsd The purpose of this modification is to revise the statement of work to reflect a within scope change, increase the contract ceiling, provide incremental funding, and authorize the contractor to acquire additional property under Subsection H.9 of the contract. Accordingly, the contract is hereby modified as reflected on page 2.
l D
eg}((g/
9311010209 930916 i
PDR CONTR t
NRC-04-91-054 PDR s provided herein, aff terms and conditions of the document referenced la item 9A of 10A.as tieretofore chanted, 36A. t4AME AND TIT LC OF CONT RAC14NG OF FICER (rype or prin#
15A. N AMC AND 18iLE OF $1GNER (Type or print /
Brenda J. Young Contracts Manager, Oak Eidge Joyce A. Fields, Contracting Officer 16C.DATE $6GN g
J SC. DATE 58GNED 368. UNIT ST AT ES OF CRI 41 RACl OR/O n ag c-orcint-<una Ottuni
\\gEp p 7 ggg3 9/16/93 gy At
<sian.<- oper.p.onupo, tens STANDARD FORM 30 (RCV.10-30-105 tiSN 754041152-60 0 Prescrit>ed by GSA namom ec resevenu e restK Atit.C F A S4 tas CF R151,2d3
l l*
l Contract No. NRC-04-91-054 Modification No. 5 Page 2 of 2 l
The Contractor and Government mutually agree as follows:
(1)
Pursuant to paragraph 4.7, Quality Assurance, of the statement of work the contractor's procedures for quality control / quality assurance are hereby incorporated (Attachment I and 2, respectively).
(2)
Section B.3, " Consideration and Obligation", is hereby modified to reflect an increase of $112,674 in the contract ceiling amount.
Accordingly, paragraphs a.,
c., and d. are modified to read as follows:
a.
The total estimated cost to the Government for full performance of this contract is $1,132,175, of which the sum of
$1,048,591 represents the estimated reimbursable cost, and of which $83,584 represents the fixed fee.
c.
The amount currently obligated by the Government with respect to this contract is $1,132,175, of which the sum of $1,048,591 represents the estimated reimbursable costs, and of which $83,584 represents the fixed fee. For further provisions on funding, see I
the LIMITATION OF COST clause incorporated in Section I.
d.
It is anticipated that the amount currently allotted will cover the cost of performance of the contract through August 4, 1994.
(3)
A summary of obligations under this contract from date of award through the effective date of this modification is as follows:
Total FY 91 Obligations 300,000 Total FY 92 Obligations 400,000 Total FY 93 Obligations 432.175 Cumulative NRC Obligations
$ 1,132,175 This modification obligates FY93 funds in the amount of $112,674.
(4)
In accordance with the contractor's revised proposal dated July 1,1993, i
Section H.9, Contractor-Acquired Property, is hereby modified to include the following equipment:
Item Description Quantity Not to Exceed Personal Computers 386DX 5 each
$ 5,805 Laserjet 4 Printer 1 each 1,369 Baytech Printing Device 1 each 353 Cables (for Baytech) 5 each 9
Adaptors 11 each 56
$ 7,592 All other terms and conditions remain unchanged.
NRC-04-91-054 Modification No. 5 l
DATA ENTRY AND QUALITY CONTROL (QC)
PROCEDURES FOR REIRS DATA ENTRY j
i The following describes a brief plan for the data entry and quality control procedures for the REIRS database. The objective is to ensure that the data is being entered accurately and efficiently within a 1% error rate for the key fields of interest. Entry and OC activities have been streamlined to maximize the efficiency of data entry while ensuring the highest level of quality.
Assumptions:
j 1.
Imts are received from the NRC pre-sorted between reactor and non-reactor submittals.
2.
Only "Old Part 20" (Form 439) termination submittals are considered here.
3.
A data " lot" contains approximately 2000-3000 records.
4.
Lots are treated as proprietary during the entire record life cycle. Records are kept under " lock and key" when not in use and access is restricted to the SAIC employees.
Data Entry Pmeess:
Lot Processing 1.
As lots are received from the NRC, the Data Entry Coordinator (DEC) logs them into the LOTLOG database. Lot boxes are marked with a lot number. Numbers contain the year and should be sequential. If a lot number is missing in sequence, alert the Ixad Analyst.
2.
Lot boxes are stored in record storage until processed. As " banker box" storage units become available, new lots are opened and transfered to these boxes. The lot number should be clearly marked on the front packet of the lot box.
3.
Data entry proceeds in lot number order. As new lots are required, lots are transfered from storage to the data entry workstations. Imts may be sorted or otherwise prepared by the data entry staff to expedite data entry. Once data entry begins on a lot, the hardcopy reports should remain in the same order they were entered for case of retrieval.
4.
After data entry a lot box is returned to storage. Storage date is recorded in LOTLOG. If the lot is selected for QA, it will remain in the drawer until QA is complete. Ifit is not needed for OA, the lot may be returned to a " mailing" box with the lot number clearly marked for storage until destruction. Destruction is scheduled at 2 years after the data has been processed completely through data entry.
i REIRS 1
July 30,1993
i i
Scheduled and actual destruction dates should be entered into LOTLOG.
Data Entry 1.
Each new lot begins data entry in the < CURRENT > work area in the REIRS data entry software. New lots cannot be started until data entry for the < CURRENT >
work area is complete and the data is stored to a database. When you are absolutely sure the current work area has been saved to a database (see DEC or Lead Analyst) you can celect < Clear Current Work Area > from the FUNCTION main menu option.
2.
Each lot is generally divided into sets of 1000 records. This speeds up the response time of the data entry software. Each sub-lot is labeled with the lot number plus A, B, and C identifiers. Therefore the first 1000 records oflot# 92-25 is lot 9225A.
3.
Begin entry in the cleared < CURRENT > work area. Enter the correct designator in the lot number field on screen. The lot number and report date carry over from one record to the next so be sure they are correct.
4.
Determine the correct license number (sometimes called the employer code or company number). If the license number is on the submittal, use that number. If not, look up the utility (company name) and use the one with the asterisk (*) next to the license number. If you are unsure, ask.
5.
For each new licensee, check the instructions for notes on the submittal. Use either the on-line Reference under the FUNCTION menu, or the hardcopy printout of the instructions.
e The REIRS data entry software contains on-line instructions for each reactor submittal. It is the duty of the data entry staff to review these instructions and point out any inconsistencies or wording that is unclear. Be very diligent in this matter.
We will meet about every 2 weeks to discuss problems and new submittals.
6.
Procede with data entry for the lot until you reach about 1000 records.
7.
Print the lot to diskette using the < Print to File > selection under the FUNCTION main menu. Use a new diskette or one that contains a printout from the same lot, j
Output files will not be overwritten and will be generated sequentially by adding a number to the file names. REPOR'IT)1.TXT would contain the first 1000 records and REPOR'IT)2.TXT would contain the second batch. Other files are named in the same fashion. Keep these files straight.
Write on the diskette label which files are for which lot and the number of records:
i REIRS 2
July 30,1993
4 REPORT 01 RX92-25A 1,052 records REPOR'IV2 RX92-25H 1,003 records 8.
When the records have been written to file the software will ask you to enter a lot number designator to store the records to a database. Only 6 characters are available for this designator. BE VERY CAREFUL WIIEN ENTERING TIIIS DESIGNATOR. If you are on lot RX92-25A and this is the first time you have printed the data to file, enter: X9225A. You will see a message on screen that the data are being written to file. His message assures you that this data is being written to a NEW set of files.
If this is the second printing after QC or some other reason, then you should enter the same designator that you used the first time. In this case, you will get the message that says that the file already exists and that the data are not being saved to a NEW file. This indicates to you that the data have already been saved and have just been updated without creating new files.
Do not save the same data to multiple f'les. Do not save RX92-25A data to X92-25 or RX9225 or 92-25A. Use distinct identifiers for each set but use the same identifier for the set each time you save. This is important. Ask if unsure.
9.
Once you have printed out to a file, be sure that the files are O.K. QUIT out of REIRS. Do a directory of the diskette to be sure that 4 files were created. Each file should have a size greater than zero bytes. For some data sets however, the INTERNAL file will be zero bytes since no internal data were entered. If you entered internal data and it shows zero bytes, tell Lead Analyst. Here is a typical directory for about 500 records:
EMPLOY 01.TXT 27754 o6-24-93 1:57p REPORT 01.TXT 33408 06-24-93 1:57p EXTERN 01.TXT 121658 06-24-93 1:59p INTERN 01.TXT 250 o6-24-93 1:59p You may also do a " check disk" to see if the diskette memory is o.k. Type CHKDSK B:
at the DOS prompt. If you get any " bad sector" or " bad cluster" messages, tell Lead Analyst or the computer service center. His check can be performed on any diskette or even your hard drive, but it is not necessary to run it often.
The best check is to print out the REPORT 01 file. This needs to be done anyway for the OC process. See DEC for printing instructions. Keep the printout in the lot box at the very front of the lot.
REIRS 3
July 30,1993
10.
Now that you have determined that the print to file was successful, you can clear out the < CURRENT > work area using the clear option under FUNCTION. Begin entry on a new lot in the clear area. Iet DEC know the hours spent on data entry and when you finished it, also schedule QC review through her.
OC Process The QC " Quality Control" process consists of a review of a summary record for each submittal. The summary contains most of the key fields, and therefore is a good indication of the data quality. Since direct data entry at less than 1% error is hard to maintain, this process catches most of the errors bringing our accuracy above 99%.
1.
OCis not to be performed by the person who did the data entry. Someone else must review your work.
2.
Print out the REPORT 01.TXT file to a printer. There is a Wordperfect " macro" on DECs machine to print this out in a good format. Under no circumstances should you attempt to modify this file. Also, _do not save the file in Wordperfect format.
This file will get loaded with the others on to the VAX so don't modify it.
3.
Review each line of the REPORT file printout against the hardcopy submittal. Total up the total whole body doses using a calculator and compare it to the total dose listed on the printout. Count each discrepancy between the submittal and the printout as one error. In some instances, one mistake will lead to multiple record errors. For example, if the license number changes and the entry person does not notice, several records will be wrong since the license number carries over from one record to the next. This is counted as one error, since it is systematic rather than,a typo. Any disagreements in the number of errors should be resolved between the data entry personnel and the QC reviewer. Only if absolutely necessary should you ask lead Analyst to make a " ruling".
Note: 'Ihe purpose of OC review is to find our errors before the QA process. Everything we do as a team is to achieve our goal of <1% error rate. Identifying errors in QC saves us from failing QA review. Counting errors per lot gives us some indication of the accuracy of entry. The number of errors per lot is a much more important parameter than the number of records per hour.
4.
Return the REPORT printout markup to the data entry person who enteired the lot.
Clarify any questionable entries. The data entry person then makes the corrections to the database for that lot. Select Restore under the FUNCTION menu and select the lot number. Edit records until all corrections are made. Print to file again when finished. Save the lot to the same lot designator when the program complete-s the print out (see item 8 under Data Entry above). Print out the REPORT file as before REIRS 4
July 30,1993 j
A and double check that all corrections were made. Place this printout in the front of the lot when finished.
5.
With Data Entry and QC now completed, return the lot to storage.
Backups Note: It is important to keep in mind that each time a " field" (data element, e.g. report date, ssn) is entered it is saved to your hard drive at the time the field is closed.
However, the data base file is open until you log out of REIRS. If power fails or you you re-boot, the files that are open can sometimes be corrupted. There is no way to keep this from happening short of an emergency battery (costly). When an open file is corrupted, just about anything can happen. 'Ihe most probable result is that you may lose records back until the last time you logged out of REIRS. With all this in mind, the following routines should be followed.
1.
As you proceed with data entry, you may want to back up the current work area on a periodic basis to minimize the potential for lost records (see note). You should balance the time it takes to backup data against the time it would take to re-enter records. Since a backup of the current work area does not take long, you may want to back up once (end of day) or twice (lunch and end of day) per day.
2.
Keep two diskettes handy for backup, one blank and one with the most recent backup. When you backup, place the empty diskette in the disk drive and select QUIT, BACKUP TO FLOPPY. Enter the disk drive of the blank floppy.
3.
After backup, select QUIT out of REIRS. Remove the backup diskette, date it, and put in the previous backup diskette. At the DOS prompt, erase the backup files on the old backup using this command: ERASE B:7??_ INFO.DBF 4.
The diskette with the files you just erased can now be used the next time you want to backmp.
REIRS 5
July 30,1993
NRC-04-91-054 Modification No. 5 Attachment #2 QUALITY ASSURANCE PLAN FOR REIRS DATA ENTRY The following describes a brief plan for ensuring the quality of the records in the REIRS database. The objective is to ensure that the data is being entered within a 1% error rate for the key fields of interest. The method used is a statistical sampling of the data where the frequency size depend upon the size of the lot and the past experience of previous QA activities.
Assumptions:
1.
Each record is considered an " item" or unit.
2.
An error in one of the key fields is considered one error for the item.
l 3.
The goal of the plan is to ensure that the average percentage of defective records per batch is no more than 1% for an acceptable j
confidence level.
4.
An average batch or " lot" size is between 2000 and 3000 records.
5.
Records discovered to be in error will be corrected.
6.
QA shall be performed by comparing the actual report received with the database as it was entered on the PC. This will reduce the number of false errors generated from the interaction of records on the mainframe and give a better indication of the quality with which the data has been entered.
Samplina Plan:
i Freauency:
~
Lots will be sampled at random. Consideration will also be given to ensure that the random selection of lots is relatively evenly spaced chronologically among the lots. A failed lot is defined as a lot with an error rate of greater than or equal to 1% as determined by the statistical sampling process described in this document. The percentage of lots to be sampled will be based on the following:
i 1
Percentage of Lots Percentage of Lots Which Fail to be Sampled l
Less than or equal to 2%
10%
Less than or equal to 10%
25%
Greater than 10% but less than 25%
50%
Greater than 25%
100%
Key Fields:
Certain fields are considered " key" or critical to the accuracy of the database. The following is a list of these key fields. An error in any one of the key fields is considered an error in the record and uounts as one error.
- Social Security Number
- Employer Code (License Number)
- Hire Date
- Termination Date
- Monitoring Date (dates of exposure)
- Total whole body dose for each monitoring period Sample Size:
As detailed in our previous statistical analysis, the number of records to be sampled per lot is dependent upon the lot size and the sampling history.
In i
general, the larger the lot the more records are sampled. Also, the more lot failures that are detected, the more records need to be sampled.
Acceptance Number "c":
The number of acceptable errors per sample is a function of the acceptable error rate, the sample size, and the confidence level. The acceptance value is noted as "c" below. This is the acceptable number of errors that can be detected among the key fields in a sample.
For example, if "c" is 4, then 4 errors detected would still pass; 5 errors would mean the let fails.
Confidence level:
1 The confidence level is a function of the past sampling history. We will generally be using a 90% confidence level for sampling, which is considered
" Level II" sampling.. This means that there is a 90% probability of acceptin'g a sample with only 1% error, a 50% probability for 2% error, and a 10%
probability of around 3% error.
Each time 5 lots in a row pass, we may drop down one level in sample size. Each time one lot fails, we need to move up to the next higher level of sampling size.
2
The following table shows the sampling plan statistics for an average lot of 2000 - 3000 records:
Level Sample Size "c"
I 125 3
II 235 4
III 315 7
Process:
1.
Select the lot to be QA'd.
Lots are selected at random among the non-QA'd lots until 25% of the lots have been QA'd for the lot calendar year.
If our failure rate exceeds 1 in 10 (10%), a larger number of lots need to be QA'd. Certain lots have already been selected for QA.
Process these until all selected lots are processed.
2.
Pull the required sample set from the lot. This involves going to storage and locating the desired lot, randomly selecting 235 (see sample size) records from the lot.
Be sure to get a few records from each company.
It is also very useful to pull the records in order from the front of the lot to the back and to keep the records in this order.
Some lots that have already been selected for QA have a red folder in the front with records already selected.
Records pulled for QA should 1
always be kept in the red folders with the lot number marked on it.
When finished with this folder, it should be returned to the lot box in storage.
3.
Check the LOTLOG to see who entered the lot. This individual's PC should contain the data entry database for that lot.
If you cannot locate the lot database using RESTORE in the REIRS program, it has been, backed up to tape or diskette.
See Derek for further assistance.
4.
Use the QA PRINT module in REIRS to print out the records for the sample set.
If the lot has been divided into more than one set (A, B, C, etc.)
then you will have to do this for each set. Go through the sample set and mark each individual in the database for a QA printout. When you have marked each individual, print the records to diskette. Print the file using Wordperfect line printer font.
5.
Compare each hard copy report to the QA printout. Mark every discrepancy found in the record using a red pen. Only count the errors in the key fields as ' total errors for the lot.
If you have any questions about an error, as Derek.
3 i
l
'J
l i
6.
Count the total number of errors for the key fields for the lot.
If the total exceeds the number of "c" for the sample size, the lot has failed.
If it is less than or equal to "c", it passes.
7.
In some cases _ you may find a " systematic" error.
For example, all the license numbers for a batch may be entered wrong due to a license number change or a misreading of the licensee listing.
If a systematic error is detected, stop the QA processing. Do a complete analysis of the records affected and determine the exact nature and scope of the problem. Report the problem to Derek. We can adjust the lot to correct for the systematic error. Once the error is fixed, begin the QA process from the beginning by pulling a new sample set and starting from step 2.
Document the systematic error and we will bring it up at the next meeting.
8.
Report the status (pass / fail) of each lot to Derek.
If a lot passes, return the QA folder with a copy of the QA printout to the lot box in storage. Give the original QA printouts (with red pen markups) to Derek.
Every error identified in QA review will be corrected on the VAX.
If 5 lots in a row pass, begin sampling at the next lower level as described above under Sampling Plan (e.g. Level II -> Level I).
9.
If a lot fails QA, a complete review of the entire lot needs to be performed. Use the QA printout module for each set of records in the lot.
Review every record in the lot as described in 5.
When finished,
)
count the total number of errors in key fields and report this to Derek.
1 This will be used to determine the true error rate in the entire lot.
Return the QA markup for the lot to Derek.
Each error will be corrected on the VAX.
Paragraph 4.7 of the statement of work requires any work performed at this point to be accomplished at contractor's expense.
Note: For a full QA printout, the file generated by REMIT will be large'.
Be sure there is enough room on the diskette. Use a blank 1.4M floppy.
4 f.lso, if a lot fails, the QA reviewer should proceed to the next higher sampling level as described above under Sampling Plan (e.g. Level - >
Level III).
4
.