API RP 581 - 3rd Ed.2016 - Add.2-2020 - Risk-Based Inspection Methodology

RISK-BASED INSPECTION METHODOLOGY, PART 5—SPECIAL EQUIPMENT 5-61
The result of the procedure outlined above will be a modified characteristic life, η mod , as defined in
Equation (5.111).
η
mod = Fs ⋅Fenv ⋅ ηdef
(5.111)
The modified characteristic life, η mod , is updated based on the PRD’s inspection history, identical to the
method presented in Section 6.2.4 i) for the failure to open case. This updated characteristic life, η upd , is
then used to calculate the probability of leakage for the specific PRD in accordance with Equation (5.112).
d) STEP 3.4—The modified characteristic life, η mod , is updated based on the PRD’s inspection history,
identical to the method presented in Section 6.2.4 i) for the failure to open case. This updated
characteristic life, η upd , is then used to calculate the probability of leakage for the specific PRD in
accordance with Equation (5.112).
prd
l
P
⎡
β
⎛ t ⎞ ⎤
= 1−exp
⎢ ⎥
⎢
−⎜ ⎟
⎝η
⎥
upd
⎢
⎠
⎣ ⎥⎦
(5.112)
e) STEP 3.5—The probability of leakage needs to be adjusted based on the closeness the system is
operating near the set pressure with Equation (5.113). The set pressure factor, F set , is dependent on the
PRD type, operating pressure, P s , and set pressure, P set (see Table 6.7 for equations).
P prd prd
l P l F set
f) Categories of Service Severity
= ⋅ (5.113)
Guidance on selecting the proper service severity for the leakage case is provided in Table 6.12.
Ideally, the owner–user’s experience with a PRD in a particular service will provide guidance as to the
choice of service severity for the leakage case.
g) Default Probability of Leakage Rates vs Time in Service
An additional set of Weibull curves tracks the leakage failure case. In this case, the curves are not based
on a per demand failure rate but are taken from data of PRDs in continuous service (i.e. a continuous
demand, unlike the failure to open case, which is on a per demand basis). The data are collected in units
of failures/year and do not have to be multiplied by a demand rate. Table 6.13 provides the default PRD
probability of leakage vs time information using a Weibull function to describe the three types of service:
MILD, MODERATE, and SEVERE. These data are currently based on a limited amount of industry data
and should be supplemented by owner–user data where available.
As an example, the default cumulative failure (leakage) distribution curves for spring-loaded conventional
PRVs using the Weibull function to describe the three categories of service severity—MILD,
MODERATE, and SEVERE—are provided in Figure 6.3.
h) Default Weibull Parameters for Balanced Bellows PRVs
The Weibull parameters for the probability of leakage curve for balanced bellows PRVs provided in
Table 6.13 match the industry failure rate data. These data reflect a minor increase in the probability of
leakage compared to conventional valves.