In the petroleum system

process fault and its damage caused by the fault propagation chains have a serious effect on the oil and gas production and people's life. In this paper

from the fault causal chain perspective

a quantitative safety early warning method of fault propagation is proposed. First

the fault propagation chains is analyzed

and the GTST-DMLD (goal tree success tree

GTST; dynamic master logic diagram

DMLD) model is developed

by which the chain effect of the fault propagation behavior can be quantitatively modeled. It can be further used to evaluate the system safety state when the abnormal event happens. Based on the results of GTST-DMLD model

Markov process theory is introduced to establish a fault prognosis model

by which the fault consequence and its propagation direction can be predicted timely. The results given by the proposed method can help field operation personnel to carry out pre-active maintenance or emergency disposal. In the case study

the proposed method is applied to the atmospheric distillation unit and the vacuum furnace in a certain petrochemical company. It has been fully validated and provides a great help for the risk reduction and accident prevention in the field.