I. INTRODUCTIONA DCS (Distributed Control System) can generate a veryhigh number of alarms and events. Control engineers orcontrol room operators often cannot deal effectively withthis number of alarms.
When alarms proliferate their collective value as a toolfor diagnosis and preventing problems declines so theoverall effectiveness of the alarms system suffers withrisk of incidents and/or relevant economics losses.Today for many advanced control systems it is becomingessential to have a system to identify and eliminate allnuisance alarms generated by a DCS, a system that allowto manage the alarms and to enhance its reliability.One useful method for alarm improvement activity is Six-Sigma methodology based on teamwork.
The Six-Sigmaapproach includes the following phases:?· Define: Identify opportunities and Project scope?· Measure: Analyze current process and definedesired outcomes?· Analyze: ID root causes and proposed solutions?· Improve: Prioritize, plan & test proposedsolutions. Refine and implement solutions?· Control: Measure progress and hold the gains.Recognize team & communicate results.
Statistical tools provided by Six-Sigma methodology anda suitable system to detect all nuisance alarms generatedby DCS represent a useful guide and support to fixexisting alarm system and to establish a well managedsystem that provides to the process controllers with theappropriate information in a timely manner. In this way itwould be possible to easily identify cause of abnormalprocess conditions and restore the plant to its normaloperations.Scope to establish a well alarm management systemshould be:?· Enable and empower operating teams to managetheir plant.?· Maximize the safety avoiding hazardoussituations.?· Minimize environmental impact avoidingrelease in ambient.?· Push the processes to their optimal limits.
?· Avoid equipment failure decreasingmaintenance costs.II The Termoli Momentive PerformanceMaterials Specialties Plant.A. Process Description.
Momentive Performance Materials Termoli (Molise-Italy) plant produces chemical specialties like SilanesOrgafunctional Liquid, Urethane Additives Copolymersand Silicone Fluids Antifoams and Emulsions. All theseproducts are based on Silicon chemistry.Final applications of these specialties are in theautomotive, personal care, healthcare, electronics,construction, textile and leathers, pulp and paper,domestic applications.Raw materials and intermediates as well the finishedproducts sometime represent dangerous or toxicmaterials.Main equipment used in the several production(continuous and batch) processes are reactors, pumps,heat exchangers, distillation columns working under highvacuum, incinerator unit, boilers and waste watertreatment system. So the control system of the differentprocesses must be very effective, efficient and reliable toavoid the risk of incidents, emissions out of the law limitsand quality problems of finished products that will createproblems in final applications.
B. Automation system description.Termoli DCS system, provided by ABB, includes acontrol network connected to 15 Process Control Units(PCU). The signals from/to the production units (field)are governed by these PCU??™s.
On the same loop operatorinterface stations are connected. These stations include 12computers, called Conductors, installed in the controlroom governed by control room operators assigned to theprocess control of the production units. Other interfacesconnected to the same loop are the EWS (EngineeringWork Station called Composer and used for programmingthe control logics), LPM (Loop PerformanceManagement) and EAM (Enhanced Alarm Management).C. Pre-project situationAbout 6000 tags are defined on Termoli DCS, eitherdigital tags or analog tags. The type of analog tags couldbe temperature, pressure, level, flow rate, pH, electricalinput. The type of digital tags could be ON/OFF valve,ON/OFF switches, interlocks, etc??¦For each of these tags an alarm is defined.
An alarm isgenerated by when the values of process variablesdetected by the tags exceed the alarm limits in case ofanalog tags, or by when the tag change its state in case ofdigital tags. In this case a signal is sent on the consoles inthe control room advising by a sound the operators aboutthe process upset.If the alarm limits are not well defined or configured wecan have way too many signals sounding continuously onthe consoles of the control room so the operator is unableto track the status of the process, with risk of incident,equipment failure, process out of control and out of specproduction.
Each alarm signal is also recorded on a printer, so incase of alarms proliferation a lot of indecipherable paperis generated.The process of an alarm management on the ABBcontrol computers (Conductors) includes the followingsteps:1. An alarm sounds on DCS computer in thealarm panel due to a process upset.2. The control room operator silences the alarm.3.
The control room operator acknowledges thealarm.4. The control room operator recall the graphic(plant) where the tag alarming has beenconfigured.5. The control room operator takes the necessaryactions to restore normal process conditions.
6. A new alarm will be generated by DCS whenthe tag returns to normal.If the alarms proliferate it is evident that this processcannot be applied in the right way.D. Problem statementA large number of overall system alarms are moreintrusive than individual process alarms and arguably oflesser value to operators; thus they are unable to track thestate of the processes and production status.The number of alarms generated by DCS during acertain period of time can be measured by EAM(Enhanced Alarms Management).EAM is a powerful software tool developed by ABB bya devoted team in Genova, first of all, to replace thealarms printing system; it also provides additionalfunctions for intelligent alarms monitoring, alarmsarchiving, off-line alarms and statistical event analysis.
Adetailed description of this software is in the followingparagraph.III Project Scope and DevelopmentA. The DMAIC approachA Six-Sigma study started in May 2008 to monitor thealarms generated by Termoli Plant DCS. Using the EAMtools and DMAIC approach, for 31 days all the alarmshave been collected and analyzed statistically to establishthe process baseline.Particularly the team put the attention on the Silane-1production area that includes critical processes andequipment.
First of all the team introduced on DCS a site SpecificAlarm Philosophy as (work practice) based upon alarmlevels priority reduction from 16 to 5 as following:1. ALARM PRIORITY = 1 (red signal on theconsol):Critical COP/NEL Alarm – Critical OperatingParameter (COP) at the never exceed limit (NEL)requires immediate, predetermined action as specifiedin COP documentation or an alarm identified in anOperational Safety Standard that is needed to control arisk associated with a Major Process Hazard. Thesealarms are associated with a critical instrument.2. ALARM PRIORITY = 2 (yellow signal on theconsole)Serious Alarm – An alarm that warns of a conditionthat if not corrected may lead to personal injury,equipment damage, environmental pollution orsubstantial economic penalties.3.
ALARM PRIORITY = 3 (green signal on theconsole):Operator Guide Alarm – An alarm that warns of a nonoptimalprocess condition, and directs the operator tolook in a certain area of the process.4. ALARM PRIORITY = 8 (light blue signal onthe console):Bad quality signals arriving on DCS.5.
ALARM PRIORITY = 16 (fuchsia signal on theconsole):Return to normal, second/third levels alarms, highdeviations, etc.By applying the Six-Sigma methodology we then definedthe Performance Standards of the process as following:Project Y: number of alarms generated by DCS.Project y: number of alarms generated by DCS in Silane1 area.Unit definition: N?° of alarms generated by DCS per 10minutes.Defect Definition: N?° of alarms generated by DCS > 10alarms per 10 minutes.Operators Target: N?° of alarms generated by DCS