Chemical Engineering April 2013 - 60

Engineering Practice
Safety-Instrumented Systems:
Focus on Measurement Diagnostics
WirelessHART and other new approaches
help operators to achieve the needed safety
at the lowest lifecycle cost
Stephen Brown
DuPont Fluoroproducts
T
his article provides guidance
to the designer and operator of
a safety-instrumented system
(SIS), with an emphasis on conducting
measurements. The use of
best practices to ensure the strength
of the overall system design, and the
use of instrument diversity to combat
common causes of failure, are covered
briefly. Greater detail are provided on
the use of new technologies - in particular,
new smart-transmitter diagnostics
and digital protocols, including
wireless options.
Background
Best practice design of SIS has evolved
over the past decade, prompted by the
widespread adoption of the ANSI/
ISA-S84.01-2004 Standard, which is
itself based on IEC 61511. In their
first article on this subject [1], the
authors described how the new standard
encourages a user to change from
purely qualitative risk assessment to
quantitative risk analysis. One benefit
of the qualitative approach is that it
is prescriptive and hence simpler to
apply - the user is told what to do to
achieve safety.
One potential downside is that the
qualitative approach is designed to
produce conservative results, which
can result in over-design. The quantitative
approach is more flexible -
the user can use whatever approach
minimizes lifecycle cost while still
achieving the desired risk reduction.
The designer must substitute quantitative
data for qualitative descriptors;
for example, a dangerous event previously
described as " very likely " would
Mark Menezes
Emerson Process Management
now be assigned a probability of " 0.1
events per year. " That article guided
users through the process of finding
these numbers. This was found to be
a particular challenge for field devices
such as sensors. What is the risk that a
given transmitter might falsely report
a safe condition? While suppliers can
provide safety statistics that are " certified "
by third parties, those data are
typically derived from white papers or
laboratory analysis. Is the actual risk
greater in a " real world " installation?
Might the risk be significantly different
for two identical devices that are
in different installations? How can the
user quantify these " installed " risks?
The focus of a follow-up article by
the authors [2] was on strategies for
minimizing, rather than quantifying,
identified risks. Since users commonly
employ redundant sensors in
critical applications, special attention
was paid to identifying " common
causes, " which can impact both sensors
in a redundant system. For example, if
a user has identified that impulse line
plugging in a given installation might
cause the pressure transmitter to
falsely report a safe condition, rather
than trying to quantify that risk, a
better approach is to make the risk so
small that it no longer has a material
impact. The user was advised to employ
a strategy that aimed to improve
strength, diversity and diagnostics:
* Strength - Change the installation
by shortening and widening the impulse
lines so they don't plug
* Diversity - Rather than using a
second (redundant) differential
pressure (dP) transmitter on a given
58 CHEMICAL ENGINEERING WWW.CHE.COM APRIL 2013
Output current, mA
4mA
20mA
FIGURE 1. The transmitter characterizes
the " normal " relationship between current
and voltage at commissioning, then
alerts the user to changes during operation,
which might indicate a problem
orifice plate, use a vortex flowmeter.
Choose a vortex design that is
immune to the common cause of
plugging
* Diagnostics - Select a transmitter
that can detect that its lines are
plugged
Best practices
To maximize the strength of an overall
SIS design users should employ
the same best practices that apply to
any process measurement application,
whether used for safety or for
basic process control. Note that best
practices tend to be specific to a given
technology, and evolve over time with
technological advances. Some examples
of best practices with particular
relevance to safety applications include
ensuring that:
* The measurement uncertainty is
smaller than the safety margin. For
example, if a process is operated
within 5% of where it becomes dangerous,
the measurement must be
much better than ±5%. This sounds
obvious, but is often not the case.
While tools and methodologies exist
[3] to quantify measurement uncertainty,
many users do not routinely
employ these
!
Terminal voltage, V
http://WWW.CHE.COM
Chemical Engineering April 2013

Table of Contents for the Digital Edition of Chemical Engineering April 2013
