Banca de DEFESA: CELSO LUIZ SANTIAGO FIGUEIRA FILHO

Uma banca de DEFESA de DOUTORADO foi cadastrada pelo programa.
DISCENTE : CELSO LUIZ SANTIAGO FIGUEIRA FILHO
DATA : 20/11/2020
HORA: 14:00
LOCAL: videoconferência na plataforma RNP (sala PEI-UFBA)
TÍTULO:

Methodology for predictive monitoring of human error and its impact on the reliability of industrial equipment.

PALAVRAS-CHAVES:

Reliability. Human Factors. Socio-Technical Systems. Estimate parameters.
Probability of Errors.

PÁGINAS: 218
GRANDE ÁREA: Engenharias
ÁREA: Engenharia de Produção
RESUMO:

Systems reliability is applied for failure prediction. The results of the prediction models
include the effects of (i) the sampling process, (ii) the prediction model definition process,
and (iii) operational and environmental factors that determine intrinsic operational reliability.
The assessment of the impact of the sources that determine reliability led to the
choice to assess the impact of the human factor. The objective of this work was to develop
a methodology to continuously evaluate the probability of human error. Regarding sampling,
it was verified the difficulty in obtaining complete data, and a multivariate evaluation
by Structural Equation Modeling was developed. The results showed a low influence of
human errors in the record and a strong influence of the record structure of failures. As
investments in updating records can be high, a solution is to have samples without using
historical data. Regarding the forecasting models, a process flow was elaborated and the
complexity for the analysts was identified. A case study was used to demonstrate that any
method has intrinsic uncertainties that are not always considered. Regarding operational
reliability, it is the result of elements such as the variation of raw material and inputs,
environment, operational context, organizational structure and human factor. They were
assessed how the elements interact with each other. As a result, a low independence of the
elements, and a model in which the output was the Human Factor, directly determined by
the Organizational Factor. As the organization is decisive, one solution would be to use
the task as the center of the analysis.The participation of the human factor led to how to
assess the probability of human error, and vulnerabilities were identified in the assessment
methods. The most used methods adjust influencing factors for the context of the tasks,
and there is diversity in the nominal probabilities of error. A model was proposed to judge
the risk of failing to use drugs, and demonstrated the complexity of including a factor in
the assessment of probability. Regarding the nominal probabilities, two types of human
errors were identified, one is stochastic and the other related to the context. Given that the
context leads to variation in probability, one solution would be to estimate the probability
of error by monitoring the context. Considering the identified solutions, a methodology was
proposed to estimate the probability of error using the task analysis and the monitoring of
equipment variables. Initially, it was proposed to evaluate the frequency of deviations from
the task, indicated by the oscillations of monitored parameters. The method was applied
to a gas compressor and error probabilities were obtained. In a second study applied to a
set of pumps over a record period of 11 years, the variable was the vibration monitored at
each textit ms. It was found that reliability increased over time and success was attributed
to actions on the task. It was possible to measure the probability of human errors and a
process for continuous use was established, demonstrating the potential for applying the
method.

MEMBROS DA BANCA:
Interno - 1817597 - ADONIAS MAGDIEL SILVA FERREIRA
Externo ao Programa - 2297578 - ANASTACIO PINTO GONCALVES FILHO
Externo à Instituição - EDILSON MACHADO DE ASSIS
Externo à Instituição - ISAAC JOSE ANTONIO LUQUETTI DOS SANTOS
Externo à Instituição - MARCELO RAMOS MARTINS
Presidente - 2042153 - MARCIO ANDRE FERNANDES MARTINS