Streamline your pipeline scada alarm management in Australia with airSlate SignNow

How to create outlook signature

my name is john of salantis ceo of heuristics australia today i will demonstrate the capabilities of the app heuristics australia's learning intelligent alarm processor we will start with the definition of alarm processing in a scada context but before that a concept of scada alarms scada alarms notify a plant operator of changes in the plant or changes in the field quite often these are non-normal abnormal or critical problems that manifest in the field and are reported in scada as alarms for example an uncommanded breaker operation in the substation would result in scada alarms similarly an over pressure event in a gas station would generate an alarm on a scada it is common practice to configure secondary alarms or responding alarms for each main alarm that would occur on a scada an example of this is with a circuit breaker a relay could actually generate a secondary alarm for example a distance protection relay another example is auto-reclosure where a number of subsequent alarms occur after the first one on account of the reclosing of the breaker under normal operating conditions secondary alarms serve a very important purpose they provide additional diagnostic and confirmatory information to the operator however in emergency situations responding alarms tend to get in the way in emergencies alarm enunciation rates rise and it is possible to lose or miss important information lost in the flood of secondary alarms in addition to this studies have shown that operators suffer from information overload and responding alarms or floods of responding alarms can exacerbate this clearly high enunciation rates exacerbated by responding alarms is not desirable in an emergency situation because it can actually make the situation far worse in extreme cases operators lose the ability to effectively control the system over the years a number of techniques for alarm management and alarm processing have been developed conventional techniques include alarm timers alarm analysis or even the removal or deletion of responding alarms altogether from the system a common disadvantage of all of these techniques is the fact that they need to be maintained continually over the life of the scada the underlying plant will no doubt change over time and therefore the changes in the alarm processing systems will need to follow a significant disadvantage of deleting responding alarms altogether is that there'll be a loss of important diagnostic information under normal operating circumstances in the scada more recently intelligent alarm processes have been applied to this problem intelligent alarm processes are also known as iaps iaps are implemented as rule-based or knowledge-based systems a human programmer maintains the set of rules however as with earlier techniques the programmer will need to maintain the rule set continually over the life of the scada because the plant will change and therefore the rules will change with it the key advantage of iaps is the ease of maintenance of the rule set nevertheless it does need to be maintained manually an alternative technique with intelligent alarm processes is to implement a learning alarm processor learning alarm processes have the capability to self-configure they automatically recognize and learn about responding alarms moreover they do so with little or ideally no human intervention learning alarm processes offer the same advantages as iaps but without the manual programming steps heuristics australia has developed a learning intelligent alarm process at leap for scada based on heuristic's proprietary sequence graph algorithm the sequence graph algorithm autonomously learns about sequences of events without human intervention it can deal with complex and intertwined sequences but the most interesting thing about the sequence graph algorithm is that it is self-organizing or self-configuring this feature allows liap to be deployed without necessarily having to implement scada policies or specifics for a given site while such information can improve leaps efficiency it is not strictly necessary for liab to operate properly and most importantly leap continually updates its knowledge base and tracks changes in the underlying plant at the same time leap normally runs on its own server and interfaces with a scada using standard interfaces odbc and jdbc are supported intrinsically whilst opc may be supported via third-party middleware a number of third-party vendors provide such middleware now we'll demonstrate the app in action for this demonstration liap is interfaced to a schneider clear scada 2013 alarm or rather event data is transferred to lia via an odbc to adbc bridge for this demonstration clear scatter is set up to generate responding alarms these alarms are transferred to liab via the odbc doordbc bridge in this demonstration lee app is set up to display its results in sql however depending on the capability of the scada this could be re-imported and displayed on the scada itself here we have the clear scada with the alarm generators running a set of four alarm generating objects has been defined in this scada these alarm objects produce interleaved sequences of alarms let's look at one of these instances here is one of the alarm objects it comprises a logic block and a number of points the logic block cycles through and alarms each of the each of the points in sequence here's an example of one of the instances in operation a primary alarm has been raised a timer is counting down to the first responding alarm now the first responding alarm has triggered another timer starts and the second responding alarm is triggered this process continues until all of the responding alarms have fired they are then all reset and the cycle starts again here we see clearsky's alarm page as you can see the alarms are continually cycling all of this time leap has been receiving the alarm sequences and has been updating its knowledge base leap has been running for approximately an hour at this stage and it has acquired sufficient data to begin filtering an alternative is to bootstrap liap with a history of alarms going back several months but we haven't done this here in order to demonstrate its efficacy at learning let's count the number of alarms that have occurred at 6 14 pm today there are 40 alarms now let's see how many alarms are enunciated in liap liap has enunciated 38 therefore to suppress two now we'll leave the system for approximately an hour so that leap can gather some more information and update its knowledge base further an hour has elapsed and will now do the same comparison this time at 7 17 p.m over that minute 41 alarms have been enunciated by the alarm generator on the scada now let's see what liap has enunciated for the same minute 21 alarms this time therefore leap is suppressing approximately half the responding alarms having learnt this over the last hour i hope you have found this presentation informative for more information please visit our website dot thank you for your interest in leap and heuristics australia you