Pipeline scada alarm management for insurance industry

How to create outlook signature

[Music] Rob com back I'm with product management um whenin wonderware uh by Snider Electric I wanted to talk to you today about um the new release um recent release of alarm advisor and alarm advisor is a package to analyze your um alarms it can provide you um very valuable insights of what are the disturbing factors in your uh plant or your facility and it basically comes with um a couple of components this is the uh alarm advisor.com website where you can go and uh try the product for yourself here is also uh where I can log in um this is in guest mode if you would want to uh have a login and all that um download the alarm advisor software from wdn and at that point you it and you get a demo mode um that will allow you to have 2 and a half thousand records uh and you can connect that to InTouch system platform U so today it connects to InTouch and system platform and it retrieves the alarms from the aldb or the WW U aldb um the WW lmdb is our Legacy database um that InTouch uses and system platform um the database is is uh being um um the the logger the alarm logger feeds the data from the alarms from InTouch um it has to be in detailed mode um so a database that is in Consolidated mode does not work with this system so if you have a logger you have to set your logger up for uh detailed and not for Consolidated um and then that has to do with the type of record sets then the second database that we support is A2 lmdb right now so A2 lmdb is the U new database that came out with 2014 um that um that gets stored uh it's alarm stored not through the logger but through the engines directly and the using using the historian um so that is uh 2014 and 2014 R2 the newest one history block that we released in in 2014 or to are not supported yet by the tool we be later um so if I look at the um um typical things here is um this dashboard provides me with metrics these metrics are basically uh uh settings um in the system so I have to log in to get to those settings so let me log in um it doesn't like me right now yeah so basically if we look at that I'm going to pull up a Word document real quick these are this is the settings page um if you look at the emua um there are um, 1440 40 alarms speak per day 144 alarms um a day uh in regular mode one to 10 one or 10 alarms per 10 minutes uh one alarm is average 10 alarms is Peak um which are recommended guid guidelines ing to the emua and but they can vary by process and by by location um so the first thing in alarm advisor would probably be to set up these uh settings towards your uh facility so we have three sections here which is general kpis and systems systems is what we connect to so right now we're connecting from one data collecting from one database uh we can collect up to uh we can connect up to 10 databases into the system and there are three um um alarm advisor additions standard professional and Premiere uh so standard does one database professional does five databases and Premier does 10 databases um the clients so the amount of connections to alarm advisor so this is the client side that we are looking at now uh is unlimited so as many people that want to access the information can access the information so here's the um alarm rate that we talked about that are uh part of the emua and uh 1 per 10 minutes 6 per8 per hour 60 per hour Peak 8 per hours average so I can make changes to this if I want to and that gets reflected into the dashboard Matrix so let's take a look at the dashboard Matrix so the dashport Matrix can give you at a glance an overview of how the performance of your uh alarm subsystems are doing and if you're trending towards Target or away from Target um these dashboards can be edited and you can add widgets to it if you want um these widgets basically are alarm rate and and severity distribution so if I want to look at the alarm rate things that I would have to my availability here is uh I can go in and I can do it time based um so I can look for 10 minutes but 10 per hour per 8 hours 12 24 hours I can apply filters uh filters could be a tank farm in this case or reactor or all the reactors the filters can be built dynamically so you can build your own filters uh the things you want to see in your plant and the things you want to focus on so that can be a specific area can be a specific node can be um anything it can be only acknowledged alarms uh stuff like that only unacknowledged alarms um so the total rate rate by severity uh display choices um and then the comparison so you can do current current month year half quarter day week depending on how much data you have in your system or last um and last I can do last months last week last quarter like that and then you can compare compare to the previous uh which is it shows you the date here if I do month it's 31 to 341 which it was March and it's going to compare it to February all right so if I say current month it's going to take April and it's going to compare it to March um very easy very easy to set up so these you can set up by yourself and um so the the uh the alarm rate has uh two views you can switch I can show you total number or uh severity distribution so these can switch in mode um of what they are sh showing and then this pie chart is the severity distribution so it shows you how many how many you have in each category also dep metric um can be defined uh so ing to the uh standard is uh we put the standards numbers in here so um right now I'm 47% here 15 11 21 so the top ones I I have too many alarms in um critical high and medium um and that would be my goal to get those numbers reduced um and be more in compliance out of a severity distribution perspective um so if we go to the second part which is now um looking at the analysis the um um that was the dashboard so if we look at analysis really is um um and let me pull up the word document again typical use of these report systems to see when alarm rates Peak and how uh severity distribution changes over time so in a glance this shows you um if I go to current week uh then we get a better view this is a demo system that I start out this week so it doesn't have too much data in it right now um it's it's going to have it but it has to run for a time for a little while um so here I can see the uh total Mark alarm counts which is 00 uh 24 and of those were uh 230 uh low 86 medium 194 high and 214 critical in a real plant this would be bad because that would probably mean that my plant shuts down every 5 minutes um but and here's a list so you can always switch to a list view U these widgets tell you um kpi of that total uh rubber band selection so if if I would make it smaller like right now I have 15,000 alarms that I'm looking at but if I to go to the current day it would basically zoom in and say oh you're looking at 4 and half th000 of 15,000 alarms right so um and you can then um and this time scale changes it's now hourly so it went from a um certain view to hourly now let let's say I would want to see real quick here things that I can do here I can build filters so I can say um what's what where the alarm is acknowledged right so um basically it's acknowledged now this is a demo system so it's very badly attended there's no operator here uh it's uh um so these two times the system was acknowledged and it was uh 76 uh alarms were acknowledged uh I can also see um in here if I take that away I can say uh from a certain area and all the areas are in here so I can choose a certain reactor probably two reactors and then I can say uh was that acknowledged and I can go in here is acknowledged apply and then it will show me only by those two areas so um those were seven alarms and if I go to list View now I'm starting to drill down um and what they what what they we so brings me to the next alarm goes to frequent uh so the next report frequent report is things that happen often so it shows the seven alarms here and it shows on what days they ured so I can I can I can pick them and they show me here in the bottom what days did they occur and then I can switch to a list View and it shows me um uh the seven alarms so four reactor level in R34 react attemp and react attemp uh total duration and then uh minimum maximum duration and if they we acknowledged so in this case because we have the filter on we only see acknowledged alarms um I can also say well from all this bring me back to all of my alarms the frequent reports are used uh for um let me show you and bring back that word document for you for a minute so the frequent alarms are those alarms that occur very often and the frequent alarm report shows the Bad actors eliminating the top 10 or 20 of these can often reduce the alarm load significant when confronted with frequent alarm situations it's it's valuable to examine if the alarm limits are set properly in this case so this report will also tell how many times an operator acknowledged the particular bad act what gives an indication of responsiveness and not only that lower act rates could also indicate that operators have too many alarms um and basically um are unable to keep up with so those key metrics are presented in these widgets so these W widgets actually say okay for the current week you have um 1,900 alarms 285 alarms a day and and then um the selected alarm statistics here are 1 day and 70 hours and 1% we're fleeting and 0% of the alarms were acknowledged so basically none of the alarms were acknowledged in this case right if we go back and take the filter off we should get more alarms so now we get the top 25 you you can select how many alarms you want to see uh 25 50 40 30 10 uh by default it's 10 but you can you can increase the count and see more top top end number alarms um so that's the frequent report and that will tell you a little bit about how many times alarms are cured and which which are the Bad actors um within your facility um standing alarms are different so if we take a look at standing and standing alarms are really um uh are those alarms that are active for very long durations uh these often indicate conditions caused by faulty alarm limits by systems that might be offline so things that you could use here are plant State suppression um you could use um um shelving uh you could use um um tools to tell the system that these systems are offline and so they don't have all these standing alarms all the time um it might be also decommission but the alarms were never switched off uh a fur review of the cost will often resolve a lot of them uh standing alarms often have a high AC rate since they are always present so with frequent they come very often but standing they don't come often they're just always there these are whatever control room you go you always see this long list of acknowledged alarms and typically those are the standing alarms that um that basically are always uh on no matter what the condition of the pl and is at that point um so we talked a little bit about how we can reduce these there are two reports here um by most frequent uh so things that happen more often or standing alarms that uh basically are indicated by longest standing alarms right so um this one was on for 3 days and four hours right and the color indicates what severity it is so here's a critical alarm uh that was on for uh three days and one hour all right so this is that report and that's based on frequency standing by most frequent and the longest and then the total of uh standing alarms um which will give you the total counts which is the one in the bottom uh and you can see by time um at what time of the day week month year you had the largest amount of standing alarms so that's what this report does um the next one is um uh fleeting and fleeting is a definition basically a uh system uh setting that you can do and in this system um it's a it's um an alarm that has been on for Less uh than 30 seconds so let me take a quick look at show you fleeting um in a document so fleeting here is um typical alarms that appear and disappear frequently and are very often short of the duration too short for the operator to act upon so a lot of these alarms won't be act so typical act rates low in the low percentiles um are um probably appropriate because the operator never had a chance to act them they were gone before he could act them um criteria can be set around alarm duration chatter time so things things um they're also sometimes called chattering alarms things to do to uh get rid of these is more um put bounce and debounce timers into the system um a certain amount of time like here I'm looking for I think fleeting is defined as 30 seconds in in this system I can put it to 15 seconds or 10 seconds and then basically anything that comes and goes within 10 second is identified as a fleeting alarm so these reports you can then use um to put those bounce and debounce timers in um and basically get them eliminated that way um so there's a total one again and then uh the most frequent um uh alarms which in this case is uh reactor temp I could switch to list view again and that one uh has a count of 70 total duration was 25 minutes average duration was 21 seconds minimum four maximum 26 uh one time here it was acknowledged so as you can see that's the fleeting report then the last one that we have is consequential consequential is um basically determining that within your system if something happens uh often um um related with that might be other alarms that happen too so for example um um when this happened this is a very low proper proper ability so this is by confidence so if we look by accurence by confidence is more like a um I can say reactor temp um 98 times and 15 times this one occurred 13 times so most likely these don't have a relationship but there might be some that have a relationship between each other where you can say h um when this if this goes off 50 times and this goes off 39 times at the same time are they related do they have something to do with each other can be um because remember we can analyze multiple databases so you might have two or three databases in this system and that are shown on different terminal stations right so for an operator or for somebody that operates the facility they are um um one might be the steam utilities boiler utilities production facility packaging and they all might be impacted by one event that ripples through the system so with this you can identify these type of events if they influence things that are down line um and that can be based on confidence or urance um and those are the things that you would use uh consequential uh analysis for so this was a short introduction uh for alarm advisor um we will be um posting more videos uh there will be four instructional type of videos that we'll post later um this is more the introduction video and we'll come back with all the other detailed information and talk about it uh hope you enjoyed it and um we'll um um see you again next time bye