You are on page 1of 6

Development of an efficient alarm management package for an

industrial process plant


Roholla Khalil Arjomandi1, Karim Salahshoor2`
1.Department of Automation and Instrumentation, Refinery number 5, South Pars Gas Complex (SPGC)
Assalouyeh, I.R. IRAN.
r.kh.arjomandi@gmail.com
2. Department of Automation and Instrumentation, Petroleum University of Technology, Tehran, I.R. IRAN
salahshoor@put.ac.ir

Abstract: Alarm management is considered to be an important issue in industrial plants which is often cited as contributing
factor in industrial incidents due to alarm flood. Modern industrial automation systems usually produce more alarm data,
making their management a non-trivial task. As a result, operators are faced with a challenging environment, having more
alarms than they can effectively handle. Alarm management, in fact, seeks to identify unnecessary alarms with wrong setting
values in order to improve the current procedures for dealing with alarms. This paper aims to practically demonstrate the effect
of using alarm management in operation of an industrial process plant. For this purpose, an integrated state-based alarm
management package will be developed in a software framework using an ASPEN-based environment to simulate an industrial
debutanizer unit as case study by preserving its operational realities. The developed alarm management package is realized in
MATLAB to mimic as a third-party tool. An interface link is implemented between these two different software environments
via an OLE (Object Linking Embedding) protocol. Performance of the developed alarm management package will be
investigated in different test scenarios including normal operation, scheduled shutdown and unexpected shutdown.

Keyword: alarm management, alarm flood, software package, state-based, ASPEN, MATLAB

the effectiveness of alarm systems [5].


1. INTRODUCTION Alarm management is treated as a fast-growing and high
profile topic in the process industries [17]. Most companies
Alarm is an audible or visible means of indicating a probable
have become aware that they need to thoroughly investigate
equipment or process malfunction or abnormal condition.
and understand their alarm system performance. Alarm
Alarms place an important role in situation awareness of
management is usually necessary in a process manufacturing
operators. They are required when the operator’s attention is
environment that is controlled by an operator using a control
being demanded elsewhere, or workload is too great to
system, such as a DCS or a Programmable Logic Controller
monitor all points on his/her own. Critical process alarms
(PLC). Such a system may have hundreds of individual
indicate abnormal situation events which are potential
alarms that have probably been designed with only limited
triggers for near-miss process safety events [1].
mutual consideration of other alarms in the system. Since
Prior to the advent of DCS, alarms were predominately
humans can only do one thing at a time and can pay attention
hardwired, requiring significant thought into the placement
to a limited number of things at a time, there needs to be a
of alarms. But, introduction of the DCS has made it possible
way to ensure that alarms are presented at a rate that can be
to create alarms more easily and at a lower cost. Although
assimilated by a human operator, particularly when the plant
software alarms are convenient, the ease with which they can
is faced with a critical upset or in an unusual condition. A
be created, has led to a significant increase in the number of
well-functioning alarm system can help a process run closer
configured alarms, contributing to alarm floods even during
to its ideal operating point, leading to higher yields, reduced
minor upsets. This has raised an adverse effect on the
production costs, increased throughput, and higher quality,
operator’s ability to manage abnormal situations. As a result,
all of which add up to higher profits [2]. This paper intends
operators today are faced with more alarms than they can
to expose the effect of alarm management in industrial
effectively monitor [5]. Alarm management comprises a
process. Debutanizer unit will be simulated as a benchmark
collection of techniques, tools, standards, and procedures
problem in ASPEN software environment. Alarm
that improve the operations of process plants by improving
management software tool will be developed in MATLAB
to emulate as a real alarm management layer. The interface
of MATLAB and HYSYS is provided via OLE(Object
Linking Embedding) protocol.

978-1-4244-8738-7/11/$26.00 2011
c IEEE 1875

1
2. MOST COMMON ALARM MANAGEMENT Next, candidate alarms are rationalized which means each
PROBLEMS one is evaluated with a critical eye to justify that it meets the
requirements of being an alarm. Alarm attributes (i.e.,
Alarm problems have been known since the introduction
settings) are documented in a Master Alarm Database, which
of control room technology in the sixties. Some of the most
also records important details discussed during
important kinds of alarm problems are[3]:
rationalization - the cause, consequence, recommended
1- Badly configured alarm systems, where there are too operator response, and the time to respond for each alarm.
many unnecessary alarms in the system [3]. Alarms that pass this screening are further analyzed to define
their attributes (e.g., limit, priority, classification, and type).
2- Badly tuned alarm parameters including low and high Alarm priority should be set based on the severity of the
consequences and the time to respond.
limits, dead bands, and filters. This may lead to
Step 3: detail design
nuisance alarms when limits are too close and silent
During the detailed design phase, the information contained
alarms when limits are too wide [3]. in the Master Alarm Database (such as alarm limit and
priority) is used to configure the system. Alarm settings
3- Alarms may be irrelevant for certain operating states.
should be copied and pasted or imported from the Master
Typical examples are alarms being tuned for normal Alarm Database directly into the control system
operation and hence are irrelevant during startups, configuration to prevent configuration errors. Advanced
shutdowns, state changes, and due to disabled alarming techniques such as alarm shelving and state-based
equipments [3]. alarming are used at this stage.
Step 4: implement alarm management
4- During an upset, operators are flooded with so many During the implementation phase, the alarms in the control
alarms. In such situations, they do not know which system are put into operation. Testing is a key activity,
ones are the most important. particularly as new instrumentation and alarms are added to
the system over time or process designs changes are made.
5- Some alarms are present on the alarm display Equally important during this phase is training the operators
continuously for long periods of time. of the system so they are comfortable with it, and so they
trust it to help them do their job.
Step 5: Implement Alarm Audit and continuous
3. ALARM MANAGEMENT ACTIVITY improvement
Once alarm system is improved, it is essential to ensure that
Most plant personnel equate alarm management with
the configuration does not change over time unless the
reducing alarms; however, this is only one piece of the
changes are specifically authorized. After any change in
puzzle. The whole puzzle involves providing operators with
alarm configuration, Master Alarm Database must be
enough information to prevent abnormal situations and to
updated.
prevent the escalation of those abnormal situations that
cannot be prevented [4].
Alarm management consists of the following stages:
Step 1: Philosophy: 4. STATE- BASED ALARMING
An alarm philosophy is a comprehensive guideline for the Most alarm in a process unit pertains to the normal operation
development, implementation, and modification of alarms. state of a piece of equipment .however equipment has
This document establishes the standards for addressing all several state. A few common state examples include:
aspects of alarm management, including design, operations, • Running
and maintenance. It should also establish key performance • Not Running
benchmarks, such as the acceptable alarm load for the • Startup
operator (average number of alarms / hr). It will reflect a full • Shutdown
understanding of the alarm problem and the proper practices • Full Rate
to follow. • Half Rate
Step 2: Identification and rationalization Moreover, operating unit have different operation mode [7].
In this step, potential alarms are identified. There are many It is a well-known phenomenon that alarm showers tend to
different sources for identifying potential alarms including appear during state changes, such as startup and shutdown,
P&IDs, operating procedure reviews, process hazards and often alarms are generated from equipment, which is
analysis (PHA), HAZOPs, incident investigations, and switched off and not in operation. In themselves, such
quality reviews. Identified is a holding point for possible alarms are expected and easily understood, but the problem
alarms to be processed in the next stage. is that they may hide other alarms, from faults occurring

1876 2011 Chinese Control and Decision Conference (CCDC)


during the statechanges. If a fault occurs during a startup,
theoperators may not see the corresponding alarms,because
they were drowned out and lost in the largeShower of “usual”
alarms. State-based alarming is useful technique which
effectively suppresses those alarm floods. In state-based
alarming logic are configured to modify alarm attributes
based on the different operational states. State-based
alarming consists of two section fist state detection and
second attribute modification. at first section equipment
state and unit operation mode should be detected in sum
situation operation confirmation is necessary at second
section attribution of alarm change according implemented
to logic.The direct benefits of state-based alarm priority arethat
it gives a large reduction of presented alarms andremoves
irrelevant alarms and events. The presentation of relevant
alarmsonly, increases the trust in the alarm system.

Fig1 simulation of debutanizer unit in Aspen environment


5. CASE STUDY PA-104: pressure of tower is very important variablefor this
This section emulates implementation of a sate-based parameter low low-low, high and high-high alarm are
alarming module of alarm management in debutanizer unit. defined.
Debutanizer is selected as case study. Debutanizer is FA-101: pump will be damaged if its discharge fluid path
designed to draw off butane cut from the liquid heavy phase blocked. At this situation, operator should be aware and
produced in Depropanizer. Its simulation is done in Aspen quickly stop pump. For protection of pump flow low
environment. State-based alarming module is implementing (FAL-101) and low-low (FALL-101) alarm are defined.
in MATLAB environment. LA-103:in reflux drum D-100 level is important parameter
for this parameter four alarm low (LAL), low-low (LALL)
5.1 Process description are defined for smooth operation of downstream unit and
The schematic of debutanizer unit is illustrated in Fig.1. high (LAH) and high-high (LAHH) are defined for
Depropanizer bottom liquid is sent to the debutanizer tower preventive overflow.Table 1 contains all alarm set point.
under flow control. The input feed consists of Butane, All alarms are to be set closely to theLimits of the plant’s
Pentane, Hexane, Heptanesand, and Octane. The tower is state (“normal”, “upset”, and “shutdown”) as shown in Fig.3
equipped with both reboiler and condenser (E-100). The
reboiler uses LP saturated steam as heating medium.
Debutanizer tower bottom temperature is of approximately
135°C. The overhead gas of the tower is cooled and totally
condensed at 60°C in E-100. The liquid butane product is
then routed to the debutanizer reflux drumD-100.Part of the
liquid is pumped from the debutanizer reflux drum by means
of the reflux pump P-100 and return under flow control to
the top tray of the tower as reflux. The other part is pumped
to downstream unit.
Fig.3 effective alarm setting
5.2 Description of alarm
For awareness operators from important parameter deviation
some alarm are defined:
LA-102: level of tower is important variable. In order to
prevent overflow of liquid fluid, level alarm high (LAH) and
level alarm high- high (LAHH) are created. For maintaining
smooth operation Levelalarm low (LAL) and level alarm
low-low (LALL) are created.
FA-100: operator should be informed if steam flow decrease
or cutsuddenly. Flow alarmlow (FAL) andflow alarm
low-low (FALL) are created for this purpose.

2011 Chinese Control and Decision Conference (CCDC) 1877


Tab.1 alarm created in debutanizer unit reviewed carefully and deferent unit condition and
Alarm Set Point equipment stateshould be considered.
For Debutanizer unit three condition are considered
LALL-102 5%
1. Normal operation
LAL-102 15%
LAH-102 85% 2. Scheduled Shut down
LAHH-102 95% 3. Shut down because of problem( unexpected)
FAL-102 150 kgmole/h
FALL-102 50 kgmole/h All set point in tabl.1 are for normal operation. In normal
PALL-104 1400 KPa operation all defined alarm are enable but depending to
relevant equipment some alarm will be disabled. FA-102 is
PAL-104 1420 KPa
for protection of pump (P-100) when pump is running and
PAH-104 1480 KPa
flow of discharge decrease pump need to be stopped by
PAHH-104 1500 KPa
operator. When this pump is not running this alarm need to
FAL-101 100 kgmole/h be disabled to suppress from operator.
FALL-101 50 kgmole/h For this purpose state-based alarming module check run
LALL-103 20% status of pump and implement changing ofFA-101
LAL-103 25% setting.Fig5 show the setting of FA-101 when pump P-100
LAH-103 80% is run Fig6.show setting of FA-101 when pump P-100 is
LAHH-103 85% stopped

5.2 Implement (advanced methods) state-based


alarming
Depending upon the capabilities of the native control
system, additional third-party tools may be required to
Deliver the benefits of alarm management. Deferentmodule
of alarm management with using OPC (OLE for process
control) protocol communicates with DCS to exchange data.
OPC specifies the communication of real-time plant data
between control devices from different manufacturers.
In Fig.3 the schematic of using OPC in industrial process is
illustrated.
Alarm
OPC management
DCS server Fig5. Attribute setting FA-101 to create alarm
module

Fig.3 schematic of using alarm management module in process plant

The proposed alarm management software package is


implemented in MATLAB. Therefore, the required link is
maintained through an OLE (Object Linking and
Embedding) protocol to enable data exchange between
MATLAB and AEPEN.

ASPEN OLE MATLAB

Fig.4 schematic of using Aspen and Matlab to emulate alarm management

In state-based alarming attribute (set point) of alarm is


modified based on the operating state of the plant or a piece
of equipment. As described in last section in debutanizer
totally 16 alarms were defined. For implementing Fig6. Attribute setting FA-101 to create alarm
state-based alarming the detail of debutanizer need to be

1878 2011 Chinese Control and Decision Conference (CCDC)


In scheduled shutdown, operator closes HV-101 by setting
output of HIC-101 to 0%. (HIC-101 is not controller it's
function is manual loader) at this situation operator is aware
of what is going in unit and as shutdown is by plane no need
to any alarms activated so all alarm are disabled.
In unexpected shutdown all alarm need to be enable to
inform operator if it variable deviate from alarm setting.
For identifying scheduled shutdown State-based alarming
module checkoutput value of HIC-101 if it decrease bellow
5% it conclude scheduled shutdown and disable all alarm.
In Tab.2, settings of alarms during scheduled shutdown are
shown.
Tabl.2 changing attribute of alarm according to unit condition

Set Point Fig.7 all alarm activated without alarm management


Alarm Scheduled
shutdown
shutdown
LALL-102 5% 0%
LAL-102 15% 0%
LAH-102 85% 100%
LAHH-102 95% 100%
FAL-102 150 kgmole/h 0 kgmole/h
FALL-102 50 kgmole/h 0 kgmole/h
PALL-104 1400 KPa 1350 KPa Fig.8 all alarm activated in presence of alarm management
PAL-104 1420 KPa 1350 KPa
Figure7 show alarm which are logged in absence of alarm
PAH-104 1480 KPa 1600 KPa
management.
PAHH-104 1500 KPa 1600 KPa
In 20 minutes the pump was stopped by the operator thus
FAL-101 100 kgmole/h 0 kgmole/h
FAL-101and FALL-101 activated. In 30 minutes pump run
FALL-101 50 kgmole/h 0 kgmole/h and flow increase. In 35 minute two valve in discharge of
LALL-103 20% 0% pump failed so the discharge fluid path is blocked and flow
LAL-103 25% 0% decrease thus flow alarm FALL-101 activated. In 45 minute
LAH-103 80% 100% valve problem rectified and flow increase and alarm
LAHH-103 85% 100% despaired.
In 50 minute, HV-101 failed and input feed decreased to 0
kgmole/h. This event is as unexpected shutdown after that
For investigate result of our package the simulated plant in FALL-101, PAL-104, PALL-104, LAL-102 and LALL-102
ASPEN environment run for 150 minutes in two activated to warning operator. In 70minute HV-101 open
situations, first, without alarm management and second and unit come to normal operation. In unit 120 scheduled
with state-based alarming module of alarm management. shutdowns happened at this time operator set output of
In each situation this event happened: HIC-101 to 0% and close HV-101. Like unexpected
shutdown in before stage all alarm activated.
1- in 20 minute operator stop pump P-100 by plane Figure 8 show all alarm activated in presence of alarm
2- in 30 minute operator run pump P-100 management. State-based alarming modules detect unit
3- in 35 minute two valve (FV-100 , FV-102) fail so condition and disable irrelevant alarms. In20 minute when
pump discharge close operator stop pump P-100 FALL-101 is disabled.
4- in 45 minute two valve come to normal situation In35minute when pump discharge fluid path is blocked
5- in 50 minute unit shut down because of HV-101 FAL-101 and FALL-101 activated to warning operator and
fail(unexpected shutdown) protect pump P-100. In 50 minute when input feed valve
6- in 70 minute HV-101 come to normal situation and failed and unexpected shutdown happened all alarm
open 100% and unit come back to normal operation. activated to inform operator but in 120 minute when
7- in 120minute operator shut down unit bye closing scheduled shut down happened operator is aware what is
HV-101 0% and open Valve FLARE.(scheduled going in unit so all alarm are disable to prevent alarm flood.
shutdown) Comparing result of alarm logging in Fig.7 and Fig.8 it is
obvious that alarm management eliminates irrelevant alarms
Like distributed control system (DCS), ASPEN log all of by detecting unit condition.
activated alarm in separated page.

2011 Chinese Control and Decision Conference (CCDC) 1879


6. CONCLUSION

Nowadays, alarm management is considered as an


undeniable part of control system in modern process plants.
This paper addresses benefits of using alarm management in
process industry. For this purpose, a real industrial
debutanizer plant was simulated in ASPEN as a standard
simulation package to preserve the operational realities of
the plant dynamics. Then, an integrated alarm management
software package was developed in MATLAB to mimic as a
third party tool. To provide an interface link between
MATLAB and HYSYS, an effective interface was
maintained via an OLE (Object Linking Embedding)
protocol to emulate as OPC in real DCS-based cases.
Performance of the developed alarm management package
was investigated under use diverse test scenarios. It was
demonstrated that activations of the individual alarms could
be intelligently handled according to specific categories of
equipments and unit condition alarms through enabled and
disabled alarms. Typical benefits of the improved alarm
management include reduced incidents, shutdowns,
improved process control, improved safety, faster incident
reviews and more efficient maintenance.

References
[1] Mauricio Moreno Santos, Bárbara Sá. Alarm Management Program:
Implementation Experience in aPetrochemical Company
[2] Siemens Industry white paper. Setting a new standard in alarm
management
[3] Jan Eric Larsson, BengtÖhman, Antonio Calzada, and Joseph DeBor.
New Solutions for Alarm Problems
[4] Invensys Systems, White Paper. Why Alarm Management is Required
in Modern Plants?
[5] D. Shook PhD.Matrikon Inc. Alarm Management What, Why, Who and
How?. July 2007
[7] Bill R. Hollifield, Eddie Habibi. Alarm Management: Seven Effective
Methods for Optimum Performance.ISA-Instrumentation,Systems,and
Automation Society.ISBN-13:978-1-934394-00-7
[8] Aspen HYSYS Customization Guide

1880 2011 Chinese Control and Decision Conference (CCDC)

You might also like