Using Human Performance Analysis to Eliminate Errors Presented September 13, 2006 DOE Integrated Saf

1
Using Human Performance Analysis to Eliminate
ErrorsPresented September 13, 2006DOE
Integrated Safety Management Best Practices
Workshop

• Jim HummerConfiguration ManagementBechtel
  National, Inc.Hanford Waste Treatment and
  Immobilization Plant

2
WTP is the Solution to Hanford Tank Waste Cleanup

• Legacy waste from WWII and Cold War production
• 53 million gallons of radioactive and chemical
  waste
• Waste is stored in 177 underground storage tanks
• 67 of the tanks have leaked one million gallons
  of waste
• Left unchecked, the waste could reach the
  Columbia River

WTP is fundamental to cleaning up nuclear waste
See www.waste2glass.com for additional
information on WTP
3
WTP is DOEs Largest Capital Procurement

• Located in the heart of the DOE Hanford Site
• Three major nuclear facilities for conditioning
  and vitrifying the waste
• An analytical laboratory
• Electric, steam, water, and air utilities
• Operations and maintenance buildings

See www.waste2glass.com for WTP Facility Fact
Sheets
4
(No Transcript)
5
Software Applications Streamline Production

• Several integrated software applications share
  information
• Engineering
• Procurement
• Construction
• Maintenance
• Business Services

6
Component Information Starts in CIS
The Component Information System is a database
application fully integrated with CAD design
software to

• Create and assign unique component identification
  numbers to equipment, valves, pipelines, and
  in-line components
• Associate quality, safety, and technical data
  with the components
• Establish component-to-document relationships
• CIS did not exist before the WTP Project.

7
CIS Error Rates are Low, but Expectations are High

• Our target is achieve zero errors in critical
  data fields
• Over 71 thousand components are managed in CIS
• Over 2.9 million data fields are associated with
  these components
• Information is exported to eight downstream
  applications
• Critical data are monitored and action taken
  weekly
• When the error rate gt1, additional corrective
  action is initiated.

An error is any data field that is incorrect,
incomplete, or inconsistent with issued design
documents. Critical data are associated with
safety, environmental compliance, and quality
8
Eliminating Errors Calls for a New Approach

• Significant error reduction was realized from
• Modifying work processes
• Revising procedures
• Software modifications
• Additional error reduction is a management
  objective
• HPI Human performance improvement is the tool

9
HPI Provides the Insight to Eliminate Errors

• Corrective action is used to fix errors
• HPI is used to fix the conditions that cause
  errors
• The CIS HPI uses a Behavioral Engineering Model
  (BEM) diagnostic tool recommended by the Society
  of Human Resource Management (SHRM)

Higher
Leveraging the Solution (Adapted from ISPI,
2001, p. 6.3).
Lower
10
BEM Helped Define our Analysis Process

• Key activities of the HPI analysis include
• Validate that problems are the consequence of
  human performance errors
• Develop and tailor lines of inquiry to identify
  the conditions that cause those errors
• Interview the users of the software applications
• Consolidate conditions identified in interviews
• Analyze for BEM system factors - information,
  resources, and incentives, and define corrective
  actions

11
Validate Conditions are Human Performance Errors

• Adverse condition reports were reviewed to
  determine if errors were due to human
  performance, the work process, or software. The
  analysis team found examples of human error

• One skill-based error
• Several rule-based errors
• Several knowledge-based errors

Errors committed ranged from inattention to
detail to necessary violations to get the job
done in spite of work process and software
limitations.
12
Tailor Lines of Inquiry to the User

• Lines of inquiry were designed to derive
  information that could be analyzed to determine

• What working steps are followed
• What knowledge is required
• What expectations are communicated to the user
• What influences may be present in the work
  environment

Seeking to understand what those who use CIS
understand when doing their work was the goal for
discovery.
13
Find the Right People to Interview

• Fourteen users were interviewed
• Interview candidates were selected from all CIS
  users
• User picked to vertically cut through the work
  flow process
• Users represented both good and bad performers
• The number of interviewees was based on
  ANSI/ASQC Z1.4 to obtain a statistically valid
  sample size

14
Identify the Conditions that Cause Errors

• Conditions identified from interviews were
  combined and consolidated to identify less than
  adequate system factor. The BEM analysis
  determined

• 1. Information available to CIS users is
  inadequate
• Users could not identify what procedural
  requirements apply to CIS
• Shared work processes are not adequately defined
• Procedure do not adequately address the role of
  automation
• Users are not always aware of procedure changes

15
Identify the Conditions that Cause Errors
(continued)

• 2. Resources available to CIS users is inadequate
  
• Users do not understand downstream use of CIS
  information
• The CIS users guide is not kept up to date
• Schedule priority influences data quality
• 3. Incentives are not used to improve CIS users
  human performance to prevent errors
• Data discrepancy reports are not used to motivate
  users
• Discretionary rewards are used inconsistently to
  recognize performance

Corrective actions were developed with
responsible management to correct or eliminate
the conditions found
16
Recommendations

• Four information recommendation for procedures
  and training
• Three recommendations to improve resources
  available to the users
• Two recommendations to apply incentives in the
  work process
• Three OTHER recommendations
• One lesson learned submitted

17
Conclusion

• The goal of zero data errors continues to point
  to additional areas of improvement that cannot be
  reached by process, procedure, and software
  modifications without considering conditions that
  affect human performance.
• HPI analysis disclosed conditions, that when
  altered or eliminated, will benefit all users,
  and reduce errors in CIS.

18
Summary

• Human Performance Analysis was a success because
  of upfront planning
• 1. Tailored lines of inquiry helped extract most
  useful information
• 2. The many interviewees, and differing roles in
  the work process, reinforced what conditions
  affected performance
• 3. The Behavioral Engineering Model used helped
  structure the analysis

• The team and management are confident that
  corrective actions identified will help us
  achieve zero errors.