The validation of three Human
-THERP, HEART, and JHEDI
Written by Barry Kirwan
Presented by Alena VanWinkle
This page left blank intentionally
Human Reliability Assessment
• Human Reliability Assessment (HRA) deals
with the assessment of human error
potential in a system and usually occurs
within a quantitative risk assessment
3 Basic Functions of Human
1. The identification of human errors
2. The prediction of their likelihood
3. The reduction of their likelihood if
Why Validate an HRA?
If HRA’s could identify performance
problems and also improve
performance during the prediction
process this would be a powerful tool
used in solving human performance
“HRA purports to be able to do both things.”
Kirwan’s series of papers, in this presentation, is
concerned primarily with the valid prediction of the
likelihood or probability of human errors (known as
quantification) in the form of :
Human Error Probabilities
HEP = number of errors occurred
number of opportunities for error
The Central Tenet of HRA
Is that the HEP estimation process must
be reasonably accurate or
at least conservative.
If the HEP is not accurate, or is
“Risk may be under-estimated,
Or the wrong errors highlighted for reduction”
The validation of HRA
techniques relies implicitly on the
collection of real HEP data.
(Computerized Operators Reliability and Error Data Base)
Is one such collection of real HEP
data that can be utilized for validation
If an HRA technique is found
valid by its author.
It must be validated accurate and
reliable by many different assessors
on different occasions.
While there are several HRAs
There has been very few validations
has been three
Human Reliability Assessors guide.
Kirwan et al, 1988
Author of THERP 1989
HRA Review by ACSNI
Advisory Committee for the Safety of Nuclear Installations, 1991
In all three reviews it was clear…
There is insufficient validations of
techniques to draw firm conclusions
on the techniques’ predictive
accuracy and consistency.
HRA techniques fall generally
into two categories
• Those that use a database.
– (e.g. THERP, JHEDI, HEART)
• Those that use expert opinions.
– (e.g. APJ, PC, SLIM, IDM)
Data based HRAs
Use a collection of generic error
These probabilities are manipulated by the assessor
to fit the context-related Performance Shaping
Factors(PSF) in the scenario being assessed.
Quasi-data-base HEPs are then formed.
Via the assessors manipulation and
interrogations of real incident data
and expert judgments.
Expert based HRAs
Uses expert judgments of particular
scenarios and PSFs to render
information into HEPs
Technique for Human Error Rate
Swain and Guttmann, 1983
This Technique uses a database of error probabilities
modified by the assessor using PSF and other considerations.
Human Error Assessment and Reduction
This Technique is based on the ergonomics literature, and uses a set
of basic error probabilities modified by the assessor using
structured PSF considerations.
Justification of Human Error Data
This Technique starts from a set of basic error descriptors
and empirically derived error probabilities, and uses
a set of PSF questions ( answered by the assessor)
to determine the HEP.
This study as a whole is
attempting to answer three
major Validation questions
What is the degree of accuracy and precision
of the three techniques?
Is the HRA technique usable by everyone or
only by seasoned HRA practitioners?
Do these techniques lead to the reliable
identification of effective error reduction
The way in which these questions
are addressed in the study, and
the answers to them, are dealt
with in the following
You made it through
The first part.