Streamline pipeline scada alarm management in affidavits with airSlate SignNow

How to create outlook signature

so today's webinar is going to be the last webinar for 2022 for Exeter and our main focus today is mainly on very small part of the alarm management life cycle is only going to be on the result of the alarm rationalization which is basically a database and then how we can use data visualization skills on that database so that we can do review on the rationalization process itself and also trying to plan for subsequent activities based on the life cycle that's laid out in the ISA 18.2 industrial standard so before I go into everything before I start everything let me introduce myself a little bit to you guys so I'm pochan I was based in Hong Kong in the Asia Pacific region you might find it a little bit strange that um you see my resume that I uh I was I actually had a bachelor in Camp biochemistry and also nma in English literary studies um believe me I couldn't find a job with these two degrees when I was back in Hong Kong sorry but on the very weird circumstances I was actually invited by Exeter to be a scribe because of my good English for a hazard and civil selection Workshop in South Korea in Seoul and for those who don't really know much about functional safety the hazard and still selection workshops are basically risk analysis workshops specifically for industrial Safety Systems we try to derive a Swiss Target from as a result of these workshops for the safety system Within These industrial sites so during the workshop I was able to basically demonstrate how versatile I was I was very quick to be able to pick up a lot of the chemical engineering Concepts and also was able to read those piping and instrumentation diagrams within two months while I was on job for that workshop and that Workshop lasted for three months in South Korea and that's why I was hired as a technical staff as a chemical technical process safety guy in Exeter and after I was hired joining Exeter I was heavily involved with a lot of reverse calculations using Markov models for specifically for safety systems and that's what we called silverification or seal calculations and because of my demonstration of my skills in math I was actually recognized and also transferred to the headquarters here in the US and after coming to the US I also get in touch and also get involved in other part of the business like the cyber security audits project for industrial control systems and also very importantly the alarm management projects and because um of the alarm management projects usually they involved a lot of data alarm data and databases I was also introduced to data science as well so that is a very briefing introduction about myself um as I've mentioned about my work that is actually the main three views of what Exeter does functional safety which means the automation the design of the automated safety system in industrial applications and alarm management all the alarms within control system we try to manage them efficiently and the cyber security for industrial control systems or industrial applications as well and exetera is basically a bunch of different Consultants we are a big consultancy agent they're scattered around the world as you can see all these different dots on the world map and our for our business are mainly focusing on three different areas specifically for the customer as well the first area is about our software that we have our own tools that we sell our own tools especially um the excelencia which is um basically a suite of different modules to help you to be compliant with IEC 61511 and IEC 61508 um for process Safety Systems within which there is a module called Silver so the silver module is basically uh going to help you to do calculations for a risk Target or risk affirmative affirmations um using Markov models so you don't really need to understand the math behind um to actually do Markov model calculations you can just use our tools to do that well if you understand the math of course you can use simply using a lot of different matrixes because that that's basically what Markov model or model is and you can use Excel to do that but for a lot of the chemical Engineers they of course the easier it would be the easier it would be to basically manage all the different activities within the management so they do prefer to use our tool for that and other than having our own tools we also do certifications for different instrumentations different devices and also personnels um so we have Personnel certifications that's cfsp and cfse um after you have finished the exams which is not easy there are a lot of people that actually fail the exam um you basically you're showing your competency in working in this functional safety field um and of course as I've mentioned we are Consultants we also provide uh expertise on different life cycle services so why do we focus on life cycles because any kind of ieciso or Isa standard when it's especially when it's related to risk there is always going to be a life cycle and that's also something that I'm going to talk about today as an introduction to alarm management as well so in order to be compliant with the standard we have to understand how the life cycle works what are the details that goes into different life cycles what are the activities or what are the different proofs that you are following that basically proving that you are following the life cycles as much so we do can we do Consulting work on that so that is the main gist of our Exeter let's go into the main course today so our main course today is about alarm rationalization if you are not familiar with alarm management you probably do not understand what even this term means so let's take a step back what exactly is alarm management the main aim for alarm management is to optimize your alarm system why do we need to do that that is because there were a lot of alarms and there are usually a lot of alarms that already exist in your systems um in the software layer because right now everything every all the alarms are basically like a software data point within your control system but back in the 50s or 60s most of the alarms are basically hardwired I actually been to a really old um uh boiler system in Canada in Saskatchewan where they still use alarms as like a flashing light with with sirens and everything is hot wine within that and they do not have like an interface to interface with that every interface would be something that's bleeping on the panel panel that you can see physically so back then because all these are hot wires so all these are actually easier to manage why would that be the case because right now everything is in the software layer and there is a false concept from the manufacturers that they think that okay more alarms means safer plants save a site so they just set everything up on default or by the manufacturers and no one actually reviewed them so that ends up with a system with so many alarms problems and that's why we have situations like more than a thousand lumps a day and within 10 minutes we have like nine alarms coming in bleeping at The Operators and they do not know what to do as we have alarm flood situations where suddenly there could be uh more than hundreds of them coming in just within a very short period of time and also very importantly there are situations that um because within the control system in most all the control systems like Delta V yokogawa they have something built in called priority the original intent of the setup of your priority is that different alarms are basically divided into like low priority medium priority or high priority and when a low priority and a medium hydrology alarm comes comes in at the same time the operator should always handle the medium Priority First so but because no one reviewed these priority settings within the control system so most of the alarms are actually set up by default as high priorities so most in that case there's a very high chance when two alarms comes in at the same time they're all both High priorities so in that case that the operators basically they don't know how to handle them and that's why that's why we need a proper system there to handle and manage these alarm systems properly and as I've mentioned um or anything that's related to risk analysis usually when it comes to standard there's a life cycle and what does life cycle means life cycles is trying to describe the life of something and here specifically in the ISA 18.2 we're describing the life of an alarm system from its birth up to its this mice the commissioning of this whole system so we are going to focus mainly on the top four activity of within a life cycle um the first thing we're going to talk about is the philosophy so the philosophy is basically um the blueprint for everything because we need to have a plan to start everything to to give birth to this whole alarm system it is a guide for all the different activities within your management cycle it should be providing goals and targets for your alarm system so that you can follow up and then try to see do assessment to see whether you have reach the proper optimization of your alarm system and also it lays out roles and responsibility and also maintenance plan for the alarm system as well so after finishing this blueprint for your alarm management system we basically go back go into the control system and then try to identify all the potential alarms that should be fairly easy why because everything is set up in default so we can basically download all the data points from your control system and those are all the potential Labs that we need to review and then we go into the most important activity in the alarm management lifecycle which is the rationalization so what exactly is rationalization so the whole flow the workflow for the rationalization is that you bring in someone who knows about alarm management who knows about these kind of system how the ISA standards mandates different activity in the systems in the whole life cycle and then basically try to help the plant decide to be compliant with the ISA standard and this person would be the facilitator for a rationalization workshops they go into the plant and then they meet up with different peoples process Engineers control Engineers project safety Engineers operators operators supervisors maintenance people instrumentation guy and maybe other kind of Engineers as well and then they discuss for every single alarm that we have identified within the system and then review every little aspect of Prior for features or settings that are related to that particular alarm for example the uh causes for the alarm what would be the consequence if the operator ignore the alarm and then let the let the situation escalate what are the actions that the operators need to do to clear this alarm and prevent the consequence what are the urgencies the time to respond to that particular alarm um what and also try to understand what severity that's related to the consequence and then from that we can derive the priorities and also we can also understand what would be the set point trigger points for these alarm as well so after knowing all these attributes for one alarm we put them into the record and then for each of the alarm we review all the different settings and all these different attributes as well and after finishing all the reviews we basically create a database and that database is termed by the ISA standard as the master alarm database it is a database of an authorized configuration of your alarm system for all the alarms in your system and that is the output for the rationalization activity and after we've finished the rationalization we go into the detailed design phase um so what exactly is this detailed design phase so the main gist of it would be activities related to the HMI interface design which is basically the UI interface for the operator when they try to try to interfere the process within the plant like how the different equipment is being shown what the valves or maybe reactors or agitators are different controllers over there so this is an example of a HMI design hm interface not a good design though so other than the HDMI design we're also going to talk we're also going to try to understand the alarm summary this way as well like how the alarms are being uh displayed and being communicated to the different operators in the plant but my main focus today is going to be the advanced alarm design so what exactly is Advanced alarm design so by the standard it means that it's some kind of a programming some kind of layers of logic our modeling to basically modify the different alarm attributes or different alarm settings in order to help with situations like alarm flood or some kind of nuisance alarm coming up at the same time so let me give you an example a very simple example what what it means by Advance alarming so one of the example that we can do with Advanced alarming is suppression by Design suppressed by Design so what we're trying to do here is basically try to suppress the alarms so um let's think about a compressor a compressor usually has a lot of different alarms associated with it um like vibration alarms flow alarms on even pressure alarms or temperature alarms and if you know about compressors usually when there's some problem all the alarms will come in at the same time because vibration would be related to like overloading of the whole compressor and when the compressor overloaded it will basically increase the bearing temperature and then and then the temperature alarm will come in at the same time and because of temperature going high the pressure our Outlet pressure will also go high so every different alarms will come in at the same time so by suppression of design we can actually do something like we basically try to identify all the alarms that's associated within for the with the particular compressor and then put them into one group when one of the alarm comes in within that particular group um one alarm within the group is triggered then we suppress the rest of the alarms within the group until that particular one alarm is basically cleared so in that case we won't have like 10 alarms coming in at the same time in a way that the operators don't really understand what actually is a root cause because usually the alarm that comes in first is the root cause for the situation and if you're interested in root causes I actually made another webinar about root cause about the I think within this year but it was several months ago you can look up in our YouTube channel and then uh have a quick uh look into that particular webinar that I made so that is the main gist and Main example of a suppression by design a kind of advanced alarming now let's go back to the life cycle so what we trying to do today is trying to basically relay the philosophy and do the rationalization and then find out what to do within the detailed design over there as you can see because rationalization is basically uh kind of a sandwich between philosophy and also uh detailed design in that sense so what we want to do is to understand what information and Analysis um that is set within the philosophy that we have done based on the data that we have we have after the rationalization the database um we do some that we can do to do the analysis to relay and also affect the detailed design that we are going to design for the advanced alarming design and uh what kind of data visualization that we can do and we that can help us with the analysis but specifically what we want to look at would be the priority distributions and we are not just looking at priority distribution across the whole plant we're also going to look at distributions within each unit and also different distribution and related alien relation with their alarm types and also we want to look at all the non-alarm class priorities and their alarm types as well I will go into detail when I showcase the visualization draft that I built for one of the refineries project I did in the past so now we go into the data part for alarms so as I mentioned the result of uh alarm rationalization is basically a big database that contains these different datas since we have different datas right here why don't we try to basically do some analysis using those datas so the usual way how we do analysis is actually try to plot charge with them we do data visualization so we have different charts we have different ways to look at data and understanding the meaning of these or implications of what these data tell us in a way that it can provide some kind of insight and data visualization is basically that we plot everything into something that can be visualized and during this process we usually employ a kind of analysis process that's very popular that's called Eda exploratory data analysis so what exactly is exploratory data analysis basically using a lot of the basic statistical Concepts like means mode or Comedians and then try to use that to basically visualize these kind of result of the data and then try to understand the different characteristics of these data so that the characteristics can actually provide some kind of meaningful patterns and when these meaningful patterns comes up we can actually try to provide some kind of models and strategies that we can use when we try to do other kind of analysis especially when we try to do predictions like using machine learning algorithms to do predictions so let's see how exactly this process is going to fit into the data science process so the data science process as we as this title suggests is mainly to deal with data so we have data from the reality we extract those data and then we try to process the data to do the data cleaning process and during the data cleaning process we try to design and do something called a feature design which I will talk about in the next slide and then after the data is cleaned we go into the Eda process within the Eda process with plot graphs and do analysis and then we can build some kind of insight so that we can pass that into different models or machine learning algorithms to do the analysis and sometimes we can even skip this process and directly do these kind of visualize report and then based on this visualization Edition we can make decisions and today I will be mainly focusing on just directly going into visualization and then making decision based on that instead of talking about machine learning process machine learning algorithm is a really big a really big process and there's so many different models with so many maps in tantel I would talk a find some time to talk about that in the future but definitely we don't have enough time today so as mentioned there's something called feature design when we do the data cleaning process so basically we during the feature design we try to combine different parameters and features into new features for example um like we can combine different alarm attributes and then basically com produce a new kind of parameters for those attributes and that is a process called feature engineering and I will give you an example when I do the visualization example for the refinery that I did so before I go into my own visualization example would like to at least introduce you to some of the visualization functions within your control system so within different control system they already have some built-in visualization functions and this is from the Delta V control system from Amazon the first histogram here is basically showing how many alarm triggers over a period of time based on that different uh priorities over there and the pie chart here is basically showing the priority distributions which basically means that um how many alarms are actually of what percentage are actually taking up the number of total alarms out there based on their different priorities out there as you can see these analysis are really actually quite I would say not exactly that primitive as a decoratory term but it's not going to be enough when you try to design some have some Advanced alarming um Advanced alarming programming situations these information is not going to be enough for that so that's why when I do visualization I would do something that's a little bit more in-depth a little bit more advanced than what's being shown uh by the um uh by from the innate program or functions from the industrial Control Systems so what I do is that I use the database Master alarm database and create some Tableau function based on that so let me show you the Tableau that I made as a draft for one of the um Refinery that I did in the past so this is the um a draft for one of the Tableau visualization I did for one of the refinery up in the Rust Belt region so as you can see it's divine it's like an infographic um for the basically for different uh stakeholders to look at for different Engineers to look at even for the management to look at as well um the main the main thing that we want to look at would be first of all as I mentioned um the priority distributions so as you can see there are three different priorities here so within Delta V system they name the priorities instead of high medium and low they name it as critical warning advisory so critical is high priority warning is medium priority and advisory is basically low priority up there and immediately you can see there are problems within this distribution because this is a result this is a resultant um priority distribution from a rationalization so all the review has been done but we are actually having more alarms than our than the targets uh than the targets that's set out within the IEC standard and that is going to be a problem um because when we get into a situation that uh we have two alarms coming in at the same time that means that there are higher chance of them coming in as high or medium priority at the same time so that is going to create stress for the operator and oh and also it's called it's also another problem is that you see the advisory is actually falling short of the target within the uh within the philosophy as well so that means that for the low priority alarms we actually falling short in terms of the percentage how much we want for the total number of uh advisory alarmed within this system so that is the total number or the total priority distribution for every alarm across the whole system now let's look at each of the subunit and units to in order to visualize the priority more efficiently I created a new prior parameter when I try to visualize them so this is called a long priority number which I named it myself so what it means that is the summation of the priorities within each unit or each subunit what does that mean so I assign a number to each of the priority up there for critical I assign number three warning I assign a number of two advisory I assign a number of one and uh and now I add everything up the number the total number of everything for each of the unit and each of the unit is basically or each of the subunits is basically being shown as one of these a bar over here as you can see and immediately we can identify three main problematic subunits out there as you can see the bars are really high and also the and also there of a different color as well so this is another skill that um you need to I guess any one that's dealing with data visualization need to know is that how we can actually highlight uh informations that basically deserves our attention in that sense and from this little round chart here we can immediately identify the three problematic units and for these three units what exactly are the problems so these three units are the Mac units the rows unit and the different boilers over there so let's look at a Mac unit first um the Mac units consists of three different subunits the Mac main the Mac flare and the Mac tank and if you are familiar with the um with any kind of a process chemical processes you are you will understand anytime the flare comes up with alarms is actually not a typical situation when the flares have alarm that means that is really serious problems so therefore our main focus would be on the Mac main here and when you look at the distribution you do realize that the Mac main the critical alarms and also the warning alarms are actually of quite High distribution over there they are taking up about 50 percent the same as the advisory alarms over there and that is going to be a problem um for the operator as I mentioned if uh two alarm comes up they would very well be in the critical order warning situation over there and this is even worse in a sense that um because usually for each of the refineries they divide operators in different groups and each group are usually going to handle each group is usually going to handle one unit at a time and for each group they're usually only one or sometimes if they have enough manpower they have two operators on duty for each of the unit and considering that there's only one or two operators when they have any kind of uh offset process offset and they have may have about 10 alarms coming in at the same time then tens alarm 10 alarms are coming in both as a critical or warning that's going to create so much stress for them in that sense so that's why we need to look into the system for the Mac system and then try to understand whether these alarms are they actually associated with a specific equipment maybe we can do what we what I mentioned before suppression by Design try to group these alarms together and then try to troubleshoot in that sense or maybe even we can just try to re-rationalize this particular unit and maybe even when we see this in situation over there um that might be a kind of indication that we need to redesign the pipeline so that we can avoid the point of using alarm to provide these kind of protection as well so these are the inside and the output of this data visualization for some kind of some kind of subsequent activities either for the detail design or some kind of review or even for other kind of design even specifically for the process itself as well now let's move on to the next unit the rose unit is actually more problematic in a way because if you look at the Rose main here they have even higher portion proportion of the critical and also warning amounts so because main would be the ones most of the alarms where they actually come from or in a way that they have a higher probability of coming from as a source for the alarms and the other problem that I see from this is that we have something called Rose Loops so Loops within Refinery or within any kind of um I would say industrial control system usually they are referring to what we call interlocks so what exactly are interlocks interlocks are automatic functions that's set up within the system that when they detect for example like a high pressure situation and then it will directly try to close some valve or stop some pump so that it can avoid that high pressure situation to further escalate but because these are automatic functions that means that usually there is no action pending from the operator because the system can take care of it itself so in that case usually these Loop would actually or interlocks would actually need to be like a notification there shouldn't be an alarm for The Operators because the operators can't do anything about them so that is something that we definitely need to go back into the system and then try to discuss with the operator about how exactly we can change that we do need to do another rationalization for this particular unit to troubleshoot that foreign to the next problematic ones which is the boiler this one is actually a little bit easier to look at because most of the alarms are actually um going to be in the boiler D so maybe the ballady has some kind of a problem with the design or maybe it's more important of the process situation over there so most of alarms are going to be the warning alarms so we basically we go into the unit and then try to review whether we can group some of the alarms together or whether we need to redo the rationalization again so that is as you can see these kind of visualization and the kind of a story that we can gain or Insight we can gain from the data visualization when we try to create a new feature and then we try to look at it in that sense so that is about the alarm priority number now let's look at the result of the different alarms that are related to the different alarm types so I created a word cloud here for any kind of alarms that are with priorities of advisory warning or critical that means high medium low um in a way that how they are related to different kind of alarms types and for any kind of work clouds the bigger the font the higher than the count the higher the number of it would be and you can see that most of the alarms are actually High alarms or low alarms so that means that they are either pressure high or pressure low or temperature high or temperature low so this is very logical because they are very important process alarms so they're usually they stay within the system as alarms and that's logical um but I do see another potential problem here which is like um the device communication device maintenance device fails and device advisory so these alarms are related to maintenance in some of the refineries the job duty of the operator is actually they look at these kind of devices um communication or kind of diagnostic alarms and then once they see these alarms they just need to alert the maintenance team for any kind of maintenance issues um so that in a way that they actually need these alarms because the operators need to do something with these kind of alarms but in some cases in some um other uh chemical plants or newer angular plants usually the maintenance team can directly assess these kind of diagnostic situations diagnostic informations so when that happens we need to consider whether we basically remove these alarms from as an alarm but rather try to substitute them or maybe change the type into something called a notifications and that's something that basically means that it's a notification to The Operators and it's not going to be appear on the alarm list so that it's not going to distract the um attention from the operator as well so these are the different alarm types that are associated with alarms now let's look at the different alarm types that are associated with notifications and right now I'm not I'm not using both word clouds here but rather using these kind of bubbles here the bigger the bubble the higher the counts of these kind of alarm types so let's first of all look at no alarms so what exactly is no alarms no alarms means that these alarms are being removed within the rationalization Workshop they are basically uh being uh disabled within the process they are deemed to be not an alarm that's useful for the process so when you look at these most of the alarms that are being uh shut down are higher alarms or lower alarms these are logical because usually higher alarms or low alarms are duplicates of high or low alarms why because the way this is basically how most of the control systems are set up when you have a high alarms like pressure high alarm you usually also have a data point that's called a high high line which is at a higher trigger point than the high alarms and since this is these are of a higher trigger points so usually they are actually a kind of a duplicate with a high alarm because they have the same cores they have the same actions so if you handle the high alarm efficiently then you actually don't need a higher alarm to do that again and do a redundancy of alarm for operators is something that the ISA standard frown upon we should not consider redundancy for human perceptions because that is only going to create a situation we call The Boy Who Cries the wolf where you just getting numb to the alarms in that sense and other than that we also see a lot of the low alarm High alarm and this deviation alarms being removed from the system as well so this is a very good indication that um this is probably the first alarm rationalization that this site has been through because if they have been through um rationalization before if or they have some kind of alarm management management already that means that they should have already removed a lot of the these nuisance alarms already but um so when they first time when this is the first time they go through the rationalizations they basically realize that okay a lot of these kind of process alarm are actually not really helping The Operators so that they basically they just remove these alarms during the rationalization process so as you can see using these chart we can actually tell another story from the rationalization process as well so that's about the no alarm alarm types now let's move on to the interlocks alarm times so in the logs as I mentioned basically means automatic functions and as you can see most of these alarms are actually High alarms at lower alarms because that means that um when the high alarms is being a law ignored or the low alarms being ignored by The Operators then the high high alarm trigger point is going to trigger some kind of automatic function so that it will allow the system to handle it just itself some kind of a like a automatic system to handle itself without the input from the operator as in like a backup in that sense and therefore these high high or low low alarm trigger points are going to line in line with those kind of automatic functions at Point as well so in that case that these are basically thresholds for interlocks and that's why they are usually being removed from the alarm list but as it that being put and substitute into the interlock um notification list in this sense so this is the basics gist of the alarm visualization at least this is the draft that I have done for one of the refinery that I've done in the past um I plan to add more visualization into the chart uh in the future but as of now these are quite sufficient because they're telling us a lot of different stories and telling us a lot of tasks and Future Works that I need to do for the whole alarm system to follow up on the rationalization process and these are can also actually show some of the key success that we have done like removing a lot of the alarms from the systems nuisance or useless alarms like we probably just a low high and deviation alarms we move about like 6 000 alarms from the system that in a way that um who can tell the uh management or the stakeholders that we have properly organized our alarm system in that sense so as an infographic it's going to tell us a lot of story in that sense so that is a very basic gist about the data visualization when it comes to the virtualization of the master alarm database the other part of data science that we usually want to go into is about machine learning models which I mentioned that I will not talk about today and also machine learning models the main focus of that on Main use for that is not just for analysis but also very much about predictions I will find some time in the future probably next year earlier I think next year to basically talk and select maybe one of the machine learning models and one of machine learning algorithm and then try to do some predictions on some safety data or alarm data and then try to understand to introduce you guys to the kind of uh Horizon that we are able to basically look at and also contribute using these kind of machine learning models to facilitate a whole safety process in that sense so if you guys are interested I really um advised you to join our course we have different courses on different aspects of functional safety and also on cyber security and also about alarms as well if you want to know more welcome to join our discussions in our LinkedIn Facebook and Twitter and I will try to answer some questions right now but um I also have an email right there and welcome also go to my LinkedIn profile and maybe ask me question over there as well um I'm also available um on the LinkedIn so if you want to write me questions I'm I'm also good with that um but right now let's see what kind of question we get so um I have a question about um about alarm rationalization itself so it's asking do you see any kind of value or visibility of conducting during the design phase of your basically of your or your system or kind of a lack of information from the plants that does not allow this kind of information so basically it's asking um whether we have any kind if we do not have any uh information as input for the alarm system then how can we actually start that in a way that I think that's what it is it um he's asking right over here so my understanding is that usually alarm rationalization is a process that you can do after you have done your Hands-On your process Hazard analysis um in a way that you basically have a lot of information like um what kind of alarms as safeguards and what kind of alarms are ipls and also because during the hazard you should have a p ID as well so you do have some of the alarms that's already within the P ID as well but um I would really advise you to see if you can or can find the control system first because there are a lot of hidden alarms within the control system so that um you can have a better picture in terms of what are the potential alarms you want to rationalize um during the rationalization process hope that answer your question so another question um so is about what kind of software we can use for alarm rationalization so exercise we do have our own software which is called seal alarm the main key um strength of civil alarm is that our civil alarm is designed ing to all the data input on data requirements based on the ISA standard so when you use that software save alarm you basically compliant with the ISA send it and they also provide a easy workflow for that so that's why we design our software in that sense um there's also um a rather it's not a question but some suggestion I think um it says Please be aware that when you do alarm reservation that you will also move alarm to Priority three to Priority two and also priority two to Priority One so that is very true yes when we do alarm rationalization we are going to review all the priority settings within the uh alarm within your alarm system and the way how we actually do that is to review the severity as well as the urgencies and then based on some Matrix we can do that if you are interested to understand how exactly we are to derive that priority welcome to a look at other webinars that are done related to alarm rationalization I believe any alarm rationalization webinars out there that's either done by me Paul Chen or maybe by Todd staffer we talked about what exactly priority means in that sense so I apologize because we don't have enough time anymore so maybe if you have further questions welcome to write me an email or write me on maybe message me on LinkedIn I'm really happy to answer you any questions and I really hope everyone can enjoy the holiday season this is a the end of year so I really also hope that everyone will have a brilliant next year and face all the new challenges and be well for 2023. thank you very much and I'll see you next time