Below is the text version for the Hydrogen Risk Assessment Models Update 2.0 Webinar video, recorded January 28, 2020.
Eric Parker, Fuel Cell Technologies Office:
Good day everyone and welcome to the U.S. Department of Energy’s Fuel Cell Technologies Office webinar series. This month we have another great presentation from Sandia National Labs on the Hydrogen Risk Assessment Models 2.0, an open source quantitative risk assessment framework. My name is Eric Parker. I provide program support within the Fuel Cell Technologies Office and I’m the organizer for today’s meeting. We’ll begin in just a second but first I have a few housekeeping items to tell you about.
This WebEx call is being recorded and will be posted online on DOE’s website and/or used internally. And by attending you consent to this recording. And all attendees will however be on mute throughout the webinar so please submit your questions via the Q&A boxes you see on the WebEx panel. Try not to use the chat box. Please use the Q&A box in the bottom right and we will cover those questions as we receive them at the end of the presentation today. And with that, I would like to introduce today’s DOE webinar host Laura Hill who is joining us at DOE headquarters. Hi Laura.
Laura Hill, Fuel Cell Technologies Office:
Hi Eric. Thank you. Good afternoon or good morning depending on what time zone you’re in. Thank you for joining us today. As Eric mentioned, my name is Laura Hill. I’m the manager for the safety codes and standards activities here in the Fuel Cell Technologies Office at DOE. One of the objectives of the safety codes and standards program is to conduct the essential R&D to enable safe and timely deployment of hydrogen technologies. One of the critical elements to meet that objective is our risk assessment activities.
So I’m very pleased today to introduce Brian Ehrhart to you all. Brian is a chemical engineer at Sandia National Laboratories. Since 2017 he has worked on the fire risk for emerging technologies team supporting technical analyses for safety codes and standards for alternative fuels particularly for hydrogen. His current and past work has focused on assessing risk for hydrogen vehicles and infrastructure, developing software codes for various fire and thermal scenarios, and working to improve the National Fire Protection Association or NFPA 2 Hydrogen Technologies Fire Safety Code. Thank you for being with us today Brian and I will hand it over to you.
Brian Ehrhart, Sandia National Laboratories:
Great. Thanks Laura. Yes. So hi everyone. As Laura said my name is Brian Ehrhart, and today I’m going to be talking to you about the Hydrogen Risk Assessment Model software, particularly the semi-recent 2.0 update that we released about six months ago through our team of developers and modelers here at Sandia National Laboratories. So in the U.S., just to kind of give some background, in the U.S., local authorities, fire marshals, and other authorities having jurisdiction refer to the NFPA 2 fire code for a standard on how to site hydrogen facilities, especially things like hydrogen refueling stations for hydrogen fuel cell electric vehicles.
In that code, there are a number of ways in which a facility can be sited and assessed to see if it’s meeting the intent and safety requirements in that fire code. The three main ways this is done, first through prescribed separation distances. So different portions of the hydrogen system have to be a certain distance, certain number of feet, certain number of meters away from different types of exposures. You can see on the right hand side of the slide here some examples of what that might be, including property lot lines, air intakes for building ventilation, that sort of thing. And these types of distances, depending on characteristics of the hydrogen system, are prescribed in the code. And so one way that a facility can be sited is just by meeting these prescribed separation distances.
If many of these requirements within the NFPA 2 code can be met except for say one or two, one alternative given in the NFPA 2 code is alternative mean and measure in which some equivalency can be demonstrated to the local authority having jurisdiction in order to demonstrate that they’re still meeting the equivalent level of safety as prescribed by those distances, but they’re meeting it in some other way―by not having that same amount of distance between exposures, maybe they’re adding something on to the system to make a hydrogen release less likely, or some other mitigation measure is in place.
Alternatively, for things that are much more unique and that wouldn’t meet a number of the prescribed requirements in the NFPA 2 code, a full performance-based design can be done. This is given in a separate chapter within the NFPA 2 code. And in this case, the individual prescribed setback distances wouldn’t be followed but rather the entire safety case for some kind of specialized facility would be demonstrated to the local authority having jurisdiction. All of the requirements and alternative methods of meeting these prescribed setback distances are based on risk informed, either prior analyses or other risk informed requirements, within the NFPA 2 code. So this risk assessment is really critical to siting hydrogen facilities in the U.S.
So overall risk, the concept of risk takes three main features or three main bits of information into account, specifically, what can go wrong or what type of scenarios could occur; how likely it is, how frequent or how probable each of those scenarios are to occur such as on an annualized per year basis; and the consequences―what exactly would happen if those events were to occur? Those three bits of information can be combined and compared either between each other or to other alternate designs or other systems in order to see which type of scenarios are really driving the risk. And what this means is that the event or the scenario, the release, might not be the most likely event. And it might not be the worst case event. It’s going to be a combination of these three pieces of information and depending on the system and depending on how these three pieces of information combine you’re going to get some combination that can reveal what actually is the highest risk type scenario. And like I said that might not be the most likely, might not be the least likely, but some combination.
As I mentioned, these, the concept of risk is used to inform and sometimes even base the requirements for a number of different fire and safety codes including the NFPA 2 hydrogen technologies code that I already talked about. This can―this was done in the past by assuming some type of representative facility, hydrogen refueling station, assessing the overall safety risk or fatality risk of that facility and comparing it to existing hazards that already exist in the world. So like gasoline refueling stations for example. Based on this a level of acceptable risk can be defined based on equivalent other risks. And specific scenarios that could occur at this representative station can then be identified and assessed. This can be done by assessing the frequency of potential leaks from this system, so the annual frequency or number of times per year that it is expected that hydrogen could be released from the system. This has been done in a number of previous analyses.
Using this type of analysis, we can assess that a certain size of leak, a certain flow rate, or a certain size relative to the pipe size of the system, will encompass a majority of the leaks. So around 98 percent of all leaks are going to be within a couple percent of the total pipe area for example. This can be presented and used by code officials in writing and applying the NFPA 2 fire code along with hydrogen release behavior models in order to estimate the effects of the leak. So you can imagine a jet flame from ignited hydrogen being released from a system that puts out some amount of heat. And that heat flux can affect people nearby or structures nearby. And that harm, that consequence can be assessed and numerically combined with the leak frequency to establish some sort of assessment of the risk of that particular scenario.
Those types of models can also be used in order to, once a particular leak size has been established as the driving risk driver for the system, that particular leak size harm, the effect of that leak, can be assessed and used to calculate these separation distances in order to specifically protect against that type of hydrogen release, what was determined to be the highest risk scenario. Very roughly speaking this type of analysis was done a number of years ago and helped define some of the gaseous separation distances within the NFPA 2 code. And so that’s why those prescribed setback distances are based on this type of risk methodology.
Out of that effort and based in large part on those types of calculations and that prior effort, the Hydrogen Risk Assessment Models, or HyRAM, software was born. It incorporates, or its main functionality is, a quantitative risk assessment or QRA methodology in order to assess both the frequency, probability of hydrogen releases as well as assess the consequences particularly for life safety for hydrogen facilities. Additionally, the software is meant to be fast running. So it can be run on any Windows PC. This doesn’t involve any sort of computational fluid dynamics that needs to be run on supercomputers or anything like that. The goal of that effort was to really get this type of assessment, these types of calculations and these models developed at Sandia and elsewhere into the hands of individuals, local authorities having jurisdiction, facility developers who might want to assess the design changes within their own facility and try to make the software easy enough to use that it can be useful for people who need it.
The most recent release of HyRAM is version 2.0. We released it around six months ago. As I said this can be installed on any Windows PC. The free download is available at hyram.sandia.gov. You can just click the link, download the executable installer. You don’t need to request a license or anything. You can just go ahead and install it on a Windows PC. Additionally, part of the HyRAM 2.0 release, we also, in a change from how we previously released the software, we’ve made the source code open source. So this is now available on github.com/sandialabs/hyram. So anyone can go to that particular code repository, look at the source code, see exactly how the calculations are done, and potentially even modify and contribute to the HyRAM software development if you’re interested in doing something like that. I’ll talk a little bit more about how the code is organized and how you might be able to contribute a little later on in the talk.
So first of all, kind of stepping through the functionality of HyRAM focusing specifically on the GUI or the graphical user interface. So this is what you get when you install the HyRAM software on a Windows PC using that downloadable installer that I mentioned. There’s a few rather simple tools used to do relevant calculations for a hydrogen system. We call this our Engineering Toolkit. This has four main types of calculations that are relatively straightforward but still can be useful. So these can allow equation of state type calculations. So estimation of density of pure hydrogen gas at certain conditions of temperature or pressure. Given a tank of hydrogen at a certain temperature, pressure, and volume can also estimate the mass of hydrogen contained within that tank at the relevant density of course.
Given a leak, given a hole in a tank, you can also estimate the mass flow rates that would occur from a leak of a hydrogen tank. Again, starting at a certain temperature and pressure and given a volume of the tank you can see if you were to have a steady state mass flow rate release you can estimate what that mass flow would be. If you want to look at the blow down of the tank, so see what the evolution of the mass flow rate would be over time, you can also estimate that. And you can do TNT mass equivalence for given the chemical energy stored in a particular mass of hydrogen as well. This might be useful for comparing to other fuels for example. The equation of state used in these calculations is used throughout HyRAM and is valid for high pressures and near ambient temperatures. So this is really focused currently on gaseous hydrogen. Not so much liquid hydrogen at the moment. We plan on soon releasing an update which will include validated models for cryogenic temperatures and liquid hydrogen. And I’ll talk more about that later on in the talk as well.
So in addition to these types of density and condition-type calculations that can be done within HyRAM we also have what we call the physics mode within HyRAM. These contain a number of hydrogen release or physics models that can be used to estimate how hydrogen can behave when released from a system. Specifically, we have a model to estimate the dispersion of a pressure-driven hydrogen release that is unignited. So a dispersion plume. You can see where the flammable concentration of hydrogen might exist, for example, in order to determine hazardous locations or safety setbacks. We also have an ignited hydrogen jet plume model. In this case you can calculate things like the temperature fields of the ignited hydrogen flame. You can also look at heat flux calculations at various distances away from the hydrogen release and use that again to estimate potentially hazardous locations or safety setback distances or whatever else is of interest for a particular system. And lastly, we have an unignited hydrogen release model which models accumulation within an indoor enclosure. So how might a layer of hydrogen develop within some sort of indoor enclosure or room. And how might ventilation affect that layer of hydrogen concentration over time? And so those are again not really focused on the frequency or probability aspects of risk. But these models are what allow us to estimate the consequence, what can actually happen should a hydrogen release occur.
And lastly, the hydrogen QRA mode is the kind of overall calculations to estimate at risk on a quantitative basis. This has a lot more kind of complex or complicated type calculations and a lot more inputs because this overall calculation really kind of wraps a lot of these previous models that I’ve discussed all together in an overall risk calculation. So overall the user can input all sorts of different information about the system under consideration, including a number of different types of components, overall system pressure, nominal pipe size. Again, those are relevant for assessing leak frequencies for different types of components and estimating the effect of a hydrogen release should one occur.
Generic probabilities and leak frequencies are provided within the HyRAM and can be used. These are based on previous analyses done and published at Sandia. However, these are changeable. If a user has more or different information about their particular system, these different probabilities and leak frequencies can be changed to better reflect a particular system. Consequence models including the hydrogen release models, different aspects of those inputs can be changed and are user editable. Additionally, different harm models, so different ways of estimating the probability of a fatality given a different level of physical harm can be used, different models can be used in different inputs to those models can be changed by the user.
Once all these inputs are in the software the overall QRA calculation can be run. This usually just takes a couple of seconds to do all the calculations and provides overall risk metrics for the system as a whole as well as more individual scenario-specific risk metrics which can be used to really see what type of scenarios are driving the overall risk for the system. Each of these overall QRA calculations can be saved locally on your computer. So if you’ve changed a bunch of inputs you don’t need to go rechange all the inputs the next time you want to run that calculation.
So this quantitative risk assessment or QRA calculation as I said really encompasses a lot of different models and includes a lot of different types of calculations. We made some changes particularly to how the leak frequencies are estimated in the most recent release of HyRAM 2.0. And so I want to give a quick overview of how these risk calculations are done, go a little bit more into the detail of how especially those leak frequencies are estimated within HyRAM in order to better describe the specific change we made and what impact that might have on calculations you might want to run within HyRAM.
So this slide shows the overall flow chart of how the risk calculation is performed. First the annual frequency for expected number of leaks per year for different size of leaks. We look at five different order of magnitude sizes of leaks relative to the cross sectional area of the pipe in terms of percentage of that area.
Using those different, five different leak sizes we can estimate the annual frequency of a leak occurring, then use different probabilities for an outcome should a leak occur. Will it be detected and isolated? Will it ignite or may it just dissipate without igniting? So there’s four different outcomes for the hydrogen release. So given those five different leak sizes and four different outcomes that combines to give 20 different scenarios that could occur for this particular system under consideration. For each of those scenarios, HyRAM then uses those physical release models to calculate and estimate the effects on nearby people, for example estimating the thermal heat flux to a person either within the facility or nearby to the facility. And again, that’s done for each of those different leak sizes that could occur.
Once that actual physical effect, the heat flux has been calculated, the harm to those people can be estimated. In particular calculating the probability of a fatality based on that physical effect that occurs from the hydrogen ignited release. Then those risk metrics can then be combined for all 20 of those scenarios in order to obtain an overall risk metric for the entire facility, including all scenarios but also to identify risk metrics for each of the individual scenarios so that a user or analyst can really see which of those scenarios is driving the overall risk, which may help inform future safety analyses or even design changes.
So to estimate the leak frequency, HyRAM uses a fault tree type setup. Essentially what this does is allow a logical process in order to estimate what the frequency of a leak might be given different ways that leak could occur. This relatively straightforward schematic shows the fault tree for a hydrogen release for random leaks for different types of equipment. So for different types of equipment like joints, valves, compressors, that sort of thing. Within HyRAM we have different leak frequency default data available for each of them. And so each of these different types of equipment can have a different leak frequency, a different, essentially a different probability of leaking per year.
Each of these individual component leak frequencies is then combined using this logical “or” gate, which essentially just says that if you have a one percent leak size from a compressor or a one percent leak size from a hose, either one of those would lead to a one percent leak size. And so they can be combined in this same way, very similar to how you might combine something like a probability by adding different probabilities together if it’s that “or” type combination. Different events may happen. Additionally, each of those individual leak components, leak frequencies per component, is then weighted by the number of components in the system, which is an input to the calculation that the user can edit. Each of the five different leak sizes then has an equivalent fault tree that looks like this for random leaks that could develop from system components.
For the larger leak size, the 100 percent leak size, we also have an additional fault tree which is specifically for fueling dispensers, so vehicle fueling dispensers only. As you can see, this is a bit more complex of a fault tree. Different types of base events could lead to a fueling dispenser failure and release of hydrogen. These can be combined using both “or” or “and” gates depending on the specific event. And these fault trees are viewable from within HyRAM itself. Instead of the individual number of components, the leak frequencies established for the dispenser failures are instead multiplied by the number of fueling demands per year, so number of times a vehicle could fuel per year, which is a function of the number of vehicles and how often they fuel. Each of those inputs is again editable by users within the software. And that fueling dispenser again just to emphasize only applies to the largest leak size, so 100 percent leak size. And then those can be combined in order to estimate an overall number of leaks per year.
So the big change we made to this type of fault tree analysis within HyRAM 2.0. Previously users were always able to edit individual component leak frequency data. So these types of numbers that would result in random leaks for all different leak sizes were always editable by users and that continues to be true. And so, if a user has better or more applicable data than what is provided by default within HyRAM they can certainly use that in order to estimate overall leaks again for each of the five different leak sizes. Additionally, the individual inputs to the dispenser failure frequencies, that separate fault tree that I mentioned, those can be editable as well. Previously these were hard coded within HyRAM. So where a user could estimate the number of vehicles that would fuel at the dispenser, they couldn’t actually edit the failure probabilities within that dispenser-specific fault tree and that is now possible.
Additionally, we have introduced more of an override to the system. So users now have the option to bypass this entire fault tree setup overall and input directly an overall annual leak frequency for each of the five different leak sizes. This can be really useful if an individual user has better information or even historical data about their own system which isn’t necessarily appliable to every system. But if they’re analyzing that system and actually can assess in some way the number of leaks that could occur per year, that type of information can now be input directly. Additionally, the type of fault trees that I showed that are within HyRAM, the structure, the logical structure of those calculations, how the different leak frequencies are combined, that type of calculation is hard coded within HyRAM. As we saw though, there are different types of fault trees. And fault trees can get much more complex in these relatively straightforward combinations.
So now with the ability to override the fault trees within HyRAM and input an annual leak frequency directly, users or analysts that are interested to do so can estimate their own fault trees and input the results into HyRAM directly. The benefit to this is that other fault tree software that’s more advanced and more configurable than HyRAM can be used. And there are both free and paid versions of different fault tree software available on the Internet. And so different fault tree software results can be input into HyRAM in order to utilize the other aspects of the HyRAM software. So the outcome probabilities and the release models that I’ll talk about next, those can still be used even if the fault trees are bypassed and a specific leak frequency is input directly. So that really provides more flexibility to the user in order to assess different types of systems which might not line up very well with some of the default assumptions made in the HyRAM fault trees and just makes it overall a lot more configurable.
Once the annual leak frequencies for each of the five different leak sizes have been established, an event tree diagram is used in order to assess what of the four different outcomes could occur for each of the five hydrogen leaks. For each of these nodes, each of these splits within the event tree, each of these is a different probability that may occur. Those probabilities are then combined with the annual leak frequencies in order to establish an annual leak frequency for each of the 20 scenarios that I talked about previously.
So given that a hydrogen release occurs, and for each of the five different leak sizes considered, the first probability assessed is the probability of detecting and isolating a leak before any negative consequences can occur, in which case the system shuts down and the leak is turned off. The probability that this would occur is an input to the calculation that the user can change within HyRAM. If that doesn’t occur, so the complementary probability of that occurring is that it not occurs, there’s a probability that the hydrogen ignites either immediately, which would result in some sort of jet fire, jet flame from a pressurized system. Or the ignition could be delayed, in which case hydrogen gas could accumulate in some way, which could lead to an overpressure type hazard. Or the hydrogen could just dissipate without igniting at all, in which case there would be again no negative consequences at least relative to ignition. Each of these probabilities is also editable within HyRAM. Currently it’s based on a prior assessment, based on the flow rate of the hydrogen release. So it’s not just a straight kind of probability number. It actually depends on the flow rate, so the ignition probabilities change for larger leaks than they do for smaller leaks.
Each of these four outcomes then results in a different type of harm. I mentioned some of the ignited jet flame models within HyRAM. The heat flux is then calculated for this jet fire and then the harm, that heat flux, is then assessed on nearby occupants of the facility in order to establish the life safety harm of that particular hydrogen release scenario. Overpressure hazards are also part of the HyRAM calculations currently. There’s not good information available by default within HyRAM. So while this calculation can be used, often these scenarios don’t end up driving the risk. Although if a user does have better information about the type of hazard these can be input in the HyRAM calculations as well if so desired.
So as I mentioned the HyRAM software is now open source. It’s written using the Python programming language and so in addition to installing the graphical user interface on a Windows PC that I mentioned, users can also access the Python code directly in order to perform additional calculations or modify the source code. So each of these different types of calculations that I discussed previously is possible from writing a Python script or a Python program or using the command line. This allows different types of calculations to be done. Perhaps you or another user might want to perform a slightly different type of calculation than the default assumptions or default risk calculations within HyRAM itself. And that sort of thing can be done using the Python code. It does require some knowledge of how to program in Python obviously.
Additionally, there are other options that we didn’t think were necessary to include in the GUI. So even additional tweaks to the calculations themselves, such as different options for numerical solvers or different ways to specify how a release might occur, those different options are also available within the Python code. So you can also download the source code and use that type of calculation directly if that would be of interest. Additionally, you can utilize the Python package of HyRAM in your own other types of calculations either by modifying the HyRAM code itself, which could be incorporated, if you kind of send it back to us, incorporated in a future release of HyRAM for everyone to use. Or you could utilize the HyRAM package within your own types of calculations. So we’re very interested in exploring future collaboration and getting input from other users and developers around the HyRAM software. So please feel free to reach out if that would be something that would be of interest.
As I’ve said before, currently the HyRAM software focuses almost exclusively on gaseous hydrogen releases, so near ambient temperatures and high pressures. Ongoing work at Sandia and elsewhere has led to the development of Python models for cryogenic releases of hydrogen, using a model we call ColdPLUME, which has been more recently validated in the laboratory scale using some of the data you see here. So showing that actual experimental releases of hydrogen can be predicted using this ColdPLUME tool, which allows this type of Python model to be used as a predictive tool, which is exactly what is needed for incorporation into HyRAM.
So in an upcoming release of HyRAM that we hope to get out soon this will include the ColdPLUME model for cryogenic releases of hydrogen. So some of those same restrictions on temperatures and pressures will be modified so that liquid hydrogen systems can be assessed more directly using the HyRAM software. So this would change the consequence modeling, the actual release of the hydrogen model within HyRAM. We’re also undergoing some work in order to assess, for the default leak frequencies available for liquid systems, assess the likelihood of those types of leak frequencies and include that as default inputs within the HyRAM software as well.
So there are a number of other models and other possibilities that could be included in future versions of HyRAM. I’ve already talked a little bit about the upcoming improvement to, or ability to handle and assess liquid systems specifically through cryogenic release behavior models. That will be outcoming here very soon in the next couple of months. In the future we also hope to incorporate better treatment of the overpressure hazard which I mentioned previously. So better treatment of both confined and unconfined overpressure hazards from ignited deflagrations for example. We’d also like to include some sort of better treatment of a flow, hydrogen flow or hydrogen flame surface interactions to see how a hydrogen release might react with different ground types, for example, or different pieces of nearby walls, buildings, or equipment that might be nearby.
We’d also like to look into pooling behavior. So especially for liquid hydrogen systems, the ability to pool on the ground has a very different release characteristic behavior than a pressurized release plume like you might get from a leak or from a vent stack for example. So the ColdPLUME model that will be incorporated in the HyRAM focuses more on that kind of plume or jet release treatment for, especially for smaller scale releases. But looking at larger scale releases is certainly of interest as well. Already mentioned the barriers walls treatment, to really assess how different types of systems and different geometries might behave with nearby barrier walls and what sort of mitigation that might provide against hydrogen releases, and a better treatment of how likely ignition is under different situations and for different types of systems would be useful.
Additionally, we plan on moving beyond hydrogen to other alternative fuels like natural gas or propane. And in the coming year we hope to release a new version of the software which includes things like compressed or liquified natural gas. I suppose at that point we won’t be able to call it HyRAM anymore since it won’t just be hydrogen, so we’ll have to come up with some other name for the expanded software suite. But of course, the ability to continue modeling hydrogen risk and releases will continue to be a major part of that effort.
So overall, HyRAM is now an open source software and Python package, an overall toolkit for assessing hydrogen behavior both within a system, releases from a system, and how likely or probable different system releases can be. We hope that these types of calculations and tools can be useful both for system designers, safety experts, code officials, and local authorities having jurisdictions in order to better inform designs and better inform code requirements and better inform siting decisions for individual systems based on the requirements from the code or an alternate assessment of the safety as may be required in some situations.
The underlying source code for HyRAM is now freely available for anyone to look at, assess, modify, or submit changes to. And we very much welcome any potential collaboration or changes that may be suggested to the source code. You can go to that github.com link in order to see where the source code resides and how you might be able to contribute. The Sandia team as well as other researchers continue to develop new models, expand the capabilities of existing models, and incorporate these expansions and improvements into the HyRAM toolkit overall so that new advances and new understanding of hydrogen systems and the behavior and safety associated with hydrogen systems is more freely available and in some sort of format which is easy for everyone to use. You don’t have to be a modeling or a Python expert or programmer to use these models.
And as part of that any sort of external use, any feedback, and any sort of collaborative development is very much welcome and encouraged. We would love to hear from you if you have anything to add. So with that, that's all I had for today. I’d like to thank you all very much for your attention. I’d like to thank Laura and Eric for letting me talk to you today about our HyRAM efforts and the more recent updates to the HyRAM software, and I’d be happy to answer any questions.
Laura Hill:
Great. Thank you, Brian. As you guys can all see on your screen if you have a question please submit it through the Q&A box and we’ll try to get to as many of those as we can. Just one quick question to start off. I think you already touched on it vaguely, but we had a question about when the liquid hydrogen update would be available. Do you have a rough timeframe for when that is?
Brian Ehrhart:
Yes. Yeah. So roughly speaking the liquid updates, the models themselves and the validation for the models at the laboratory scale, the model and the validation has been published publicly already. And so it’s more of a matter of incorporating it into the larger software such that it can talk to the rest of these calculations. It’s also involving a change to how the physical properties of hydrogen is assessed within the software. So like I mentioned, currently an equation the state is used for gaseous hydrogen, but of course that’s not as applicable for cryogenic releases. So that’s where the effort is coming in, is changing kind of how that calculation is done. As for when the model within HyRAM should be released, we’re anticipating within the next two to three months. So we’re working on incorporating that into the software now and then we’ll do a little bit of testing to make sure that everything is still working as it should and that the models are using that different physical property method correctly. And then we should be releasing it in the next two to three months.
Laura Hill:
All right. Thank you. You touched a little bit just now on some of the underlying aspects of HyRAM. If somebody wanted to find more information about how to use HyRAM and what the underlying decisions in the software are, where could they go? Can you point them in a particular direction?
Brian Ehrhart:
Yes. So the hyram.sandia.gov, on that webpage, in addition to providing links to the downloadable installer and the source code repository, we also post links to the documentation for the HyRAM software. We currently have the documentation kind of split up into two main Sandia reports, so two main documents. One that we call the user guide provides information on specifically how to use the software, especially the front end GUI. So how to make changes to different types of inputs and where those user editable inputs are incorporated into the calculation. So that’s where you’d get a little bit more information on how to actually use the software. We also have a technical reference manual or algorithm report which talks more specifically about the calculations themselves, the assumptions and models underlying the calculations. And it provides a lot more references to various works, technical reports and peer-reviewed journal articles that underpin the HyRAM analysis data and models.
And so that technical reference manual is also available on the HyRAM website and goes a little, a lot more into detail about how the calculations are performed. Within that document you can see where different models came from, what type of assessments or systems our calculations were originally based on, and why those systems were chosen. And more detailed information about the applicability of different types of calculations. So when the equation of state is valid for example. And we will include more detailed information especially about the cryogenic releases in the future updated version of HyRAM within that same type of document. So the hyram.sandia.gov website is the place to go to access all of that information.
Laura Hill:
And one last user assistance question. For those who might have code questions or need some support what is the best method for reaching you guys or is there some sort of active online forum for discussion?
Brian Ehrhart:
So there’s a few different options. So on the hyram.sandia.gov website there is contact information for both myself and others on the Sandia team that you can reach out to via email in order to ask and get some help with specific questions depending on your particular needs. Also on the DOE FCTO H2 Tools website―and Laura, I’m sure you can talk more about that if you’d like―there is also a forum specific to HyRAM that if you have kind of more general questions or just want to kind of participate in more of a community so that not everyone is asking the same questions. For example, you can kind of see what people have asked in the past. That forum is also a possibility. On the code itself if you have more kind of code-specific questions or want to kind of ask about contributing, the link on the HyRAM website to the GitHub code repository is where you can submit things like code pull requests or kind of log bugs or issues you’ve come across within the calculations. And so there’s kind of different options, each of those may or may not be easy or applicable for different situations. But there’s different ways to get ahold of us.
Laura Hill:
All right. Thanks, Brian. We’ve got a few questions more along the lines of the functionality of HyRAM. First, does HyRAM account for different leak sizes in each component?
Brian Ehrhart:
Yes. So each of the five different leak sizes is assessed for each of the components. So if you can kind of see the―if I can get to the inputs. So different data is available for each of the different component types for each of the five leak sizes. So that kind of leads to a lot of different numbers. But it also provides a lot more flexibility within the software that different components are individually editable for each of the five different leak sizes. So that really gives a lot of flexibility in terms of what exactly you want to calculate.
Laura Hill:
Great. Another question we’ve got is how do you plan on incorporating uncertainty and sensitivity analysis into HyRAM?
Brian Ehrhart:
Excellent question. Yeah. So currently HyRAM really kind of looks at the average or expected value for each of the different scenarios in order to provide overall risk metrics for each of those 20 scenarios. Of course, as the question implies, there’s a lot of potential uncertainty within each of those types of calculations. And that can be very important in assessing the safety risk for a facility. Some of the underlying information within HyRAM really doesn’t lend itself well to sensitivity analysis. The release models for example don’t necessarily calculate a probabilistic release. They really calculate a given specific release and estimate the effect of that release on nearby people for example. Other parts of the calculation, the leak frequencies for example, were estimated with sensitivity error bars essentially, sensitivity included in that.
Currently the calculation only utilizes the mean, the average leak frequency of each of those components. But you can imagine in a future release of HyRAM we might be able to incorporate a better treatment of the uncertainty. One of the reasons we haven’t done that yet to date is it really adds a lot more complexity and a lot more computational power in order to run the same types of calculations. So running a sensitivity of a thousand different scenarios is a lot harder than running, a lot more computationally intensive than running 20 scenarios for example. And so that’s kind of been in the past why we haven’t necessarily included that sensitivity treatment specifically. We try to include conservatisms where appropriate within the QRA calculation. But better treatment of uncertainty is certainly something we’d like to incorporate into a future version of HyRAM.
Laura Hill:
Great. Also another question we have kind of along those lines. Are there any plans to qualify the code to meet certain QA requirements and how?
Brian Ehrhart:
Currently we don’t have any specific plans. Each of the models within HyRAM has undergone validation in peer-reviewed literature. And those references are available through the technical reference manual report that’s on the HyRAM website. But in terms of meeting a specific QA methodology or requirement we don’t have any immediate plans to do so. If you have any specific needs or suggestions on what would be useful reaching out to either myself or Laura would certainly be helpful. We could take a look at doing that in a future release.
Laura Hill:
Great. Thanks Brian. I have a couple more questions before we wrap up. We have one, there are six notional nozzle models built into the code. Are there any recommendations on which model has the overall best agreement with experimental data?
Brian Ehrhart:
Yes. So we kind of, historically the code and the calculations have included a lot of these different types of models. This was done partially just because they were available, they were able to be used. But also certain references assume, underlying to their own validation, assume a specific notional nozzle model. The defaults that we set within the GUI of HyRAM is for the “YuceilOtugen” model. That’s generally the one we recommend and that’s why we set it as the default. Other models make different assumptions. Each of those notional nozzle models are discussed and documented within the technical reference manual, so you can do a little bit more reading there if you’re interested. But yeah, generally we set the default to use, and we would recommend, the “YuceilOtugen” notional nozzle model.
Laura Hill:
Would there be a chance that a missile object would be incorporated into the future flows flame surface interactions feature?
Brian Ehrhart:
Possibly. So any type of overpressure hazard, in addition to kind of the direct overpressure effect of a pressure vessel or of a hydrogen release that had some sort of delayed ignition, the direct pressure effect is certainly relevant and there are ways in the literature currently that that can be considered. Additionally, like you say, any sort of fragments or projectiles, kind of missiles, that come from damaged equipment or from other debris in an overpressure event also would certainly be relevant to the physical life safety harm and therefore would certainly warrant calculation into the quantitative risk assessment within HyRAM. We’re currently looking at a number of different ways in incorporating overpressure overall into HyRAM.
It’s not always perfectly confined within some sort of enclosure. So we’d like to incorporate some sort of treatment of partial confinement. That would be really helpful. In addition to incorporating both that, like I said, the direct blast way as well as any sort of projectile probability. Each of those models would need to be kind of independently looked at and assessed for inclusion within HyRAM, which is why it’s a little kind of slow going to incorporate them into the software package as is. But yes, that would certainly be something that we can and would like to look at for future HyRAM versions.
Laura Hill:
All right. Thank you, Brian. If you would just hop back to the slide that has your contact info and the download link. Thank you everybody for submitting some great questions today. And we just want to remind you where you can go to download HyRAM 2.0. And with that I’ll turn it back over to Eric.
Eric Parker:
Sure. Thanks Laura. And thank you Brian for an excellent presentation today. And I know we did get quite a few questions. We’re going to make sure to capture all of these, and if we didn’t get to yours, we’ll do our best to address it afterwards directly or incorporate it into the presentation somehow. But other than that, that does conclude the webinar for today. I’d like to thank everyone for joining and remind you that the webinar and slides will be available online in the coming week or two. And also, I encourage everyone to sign up for our monthly newsletter, which includes information and registration for upcoming webinar topics. And you can sign upon on the DOE FCTO website as well. And with that I’ll wish everyone a great rest of their week and goodbye.