Episode 21: Tidal Modeling Using HEC-RAS with Gary Brunner

[Music] do welcome to full momentum and hec raz vodcast i am your host ben carey and here joining me as always is chris goodell and for his third time gary bruner welcome to the show today gary thank you it's great to be here gary good to see you you looking nice and relaxed there like almost like you're sort of retired huh i i am semi semi-retired well that's an awesome awesome thank gary we're happy to have you here for episode 21 of today's show we got some really exciting technical topics we're going to get into some discussions on title modeling we're going to talk about the new features in version 6.2 but before we get into anything chris and i definitely want to hear about what's going on in post-retirement life well i i um retired the end of may and in fact my last act as an employee was i got to push the button that sent out the email announcing that raz 6.0 was released last thing i did on my last day of work wow 4 30 in the afternoon well i think we need to mention too because uh just to make sure everybody's aware but gary is has worked at hcc for 30 plus years 36 right leading the heck razz team for most of that time and so it was a big deal when you retired not even a year ago right right when was that he's in may or something yeah so um yeah it's like it's almost an end of an era but um you know razz will continue on without gary gary's moved on to other things yeah so i i took five months off to find myself uh no where were you now i took five months off and not worked at all because i told somebody you know literally since i started college because i worked the whole time went through college i never had more than a two-week vacation my whole entire college work life wow you know and so i thought well i should i just want to like do nothing for a while so i did nothing for a while and i started really liking it but i i did take up some new i did more fishing i was doing more exercising i took up photography i've been doing a lot of photography a lot of photos i like uh wildlife photography so i do a lot of that i've seen a lot of your pictures gary on facebook of the you like to do birds right birds ducks anything wildlife but i do seem to migrate to a lot of birds and ducks and stuff like that aren't ducks birds i mean i guess yeah speaking of speaking of birds uh chris and i had a really cool experience yesterday we were out in eastern oregon uh near now here wildlife refuge and we saw tens literally tens of thousands were they sno what were they snow docks what were they called snow geese white with black tips on the wings yeah it was pretty spectacular yeah snow geese yeah cool yeah we walked up close to him and uh an entire flock just took off at once and it was so cool watching them take off like that i'll uh to show you the video gary maybe we'll post it up here too so gary at what point in your five you said five months right yeah vacation were you like you know what i think i'm ready to start doing something again well my original plan was only to take three months off but i liked it too much so i said heck with that i'm going to take months off but then i had always planned i you know i worked only for the federal government right out of college and i thought you know i've never worked in private industry and i think i would like to try it so i decided to hang my resume out and i got a few bites klein schmidt was not one of them by the way put us on blast you knew we didn't have a chance gary so uh i i had some interviews which was cool uh first time interviewing since uh 1985 you know so i have a lot to learn about what it's like to interview now yeah uh i got fortunate i i got um three offers and i i went with a company called hdr which i think you guys know yeah and um i'm just a half-time employee so i worked 50 time which is all i wanted to do i didn't want to work full time but i work in their sacramento office and sacramento office is really nice uh it's like 200 plus people over there when when they're fully staffed and everybody's in the office from code which hasn't happened since i started working but um they got a pretty big water resources group uh hydrology hydraulics reservoirs dam safety lots of stuff going on there so it's been really interesting yeah many people know mark forest forrest he's i still haven't wrangled him onto the podcast yet but i want to do that at some point but mark's uh uh the practice lead for h hydraulics and hydrology at hdr i think that's his title and uh so and i know you and mark go back a long ways and uh i'm guessing that was a little bit instrumental in getting you over to hdr right yeah that and the the other person um was originally the reason i was going to go to hdr is because originally i was going to work for david ford and his bought out by hdr but then i a lot of people probably know that david passed away but uh he and i were good friends and uh since conception of coming to hcc because he worked at hcc when i started here and uh he's actually one of the people that interviewed me for the job so wow yeah so very cool so yeah so you went from david ford from one interview and then 25 30 years later to mark forrest in another interview yeah yeah that's it that's kind of fun so gary everything's interconnected yeah so gary since you got to private consulting i'm curious because you had a certain view of hcc whereas it's applicability it's strength its weaknesses working for hec you then go over private consulting i'm assuming you're doing a lot of work that involves projects you know tied to raz do you have any did you learn anything about raz moving into private consulting that you hadn't really thought about or you didn't really know as much about before moving over well um i kind of already knew this but it kind of driven home is one of the things that you really need to concentrate on in the private industry is you know doing things efficiently and you know looking for ways to make uh work more efficient easier so forth and and because of that a lot of people develop spreadsheets they write their own code around raz they use this they use that and uh and i always knew that that was the case at night from my view i tried to make razz more efficient like you know the whole concept of raz is a river analysis system it's all in one you don't need to buy a gis you don't need to buy a special interface you know if you want to do steady flow unstable 2d sediment transport water quality it's all right there that was really the idea from conception it just has taken 30 years to make it happen so so that's one thing is is you know the consultancy now because let's face it time and money is the is the the bottom thing in the consulting world right you know you've got a client and you said you're going to do something for x amount of dollars that next amount of time and you got to get it done that's right so now you understand the frantic phone calls i'd have to you know in the past you know on a friday and uh begging you for help right yeah i understand yeah that's where it comes from but the other thing that is what was eye-opening to me is that um there's a lot of really smart consultants out there that have developed a lot of really cool and unique tools of their own to do their work okay so that's a big eye opener too and seeing what other people have really developed to help them do what they do you know yeah and it's it's great now with the ability to script with razz and to use the hec-ras controller to help with that and now you've got scripting built into raz mapper where you can develop your own maps and share those scripts with others and so that's one thing i really enjoy about where you've taken raz is its ability to allow the user to even take it further you know if they want and you a lot of things like you said the api and yeah you know programming against that but also then the rules capability yeah perhaps for controlling gates and stuff like that and pumps and that's i've actually been using the rules more now outside of hcc than i had inside because we're doing a lot of work with um the california aqueduct actually and and all the kind of gate settings there and i was helping some young engineers write rules to control operations at aqueduct and for different purposes so that's been fun too yeah that's awesome i i'd be interested you know even like a year two years from now to get your take on being in the consulting side of things how does that shape your view of how raz works and how it's set up and what things you might have done differently having that perspective versus you know working at hec yeah i'm sure i'm gonna have some different points of view and but one of the things about hc is a lot of what happens is dictated by money yeah yeah it just is i mean you you can't really a lot of people think you know hcc is government office they just have a truckload of money they do whatever they want that is completely false you know what is dictated is what gets paid for and whether that's through r d and you have to write proposals all the time to get r d proposals and get that through or whether a district puts up the money to pay for a new feature and that's how raz really went okay and so i wrote always writing proposals all the time for r d money and calling districts and writing just to say hey wouldn't it be cool to have this new feature you guys could really use this so gary you told us you've told us you told stories about that before and i i remember them being pretty interesting so is there a is there one feature that stands out to you that razz has and it's due to a specific district put up the money to do it well there's so many i mean the reality is that the district officers in the core paid for most of res okay yeah let's say at least 50 and then r d paid for the other 50 so there's so many features i mean from pumps paid for by new orleans district to rules paid for by you know florida you know and software or management outside the core also and you know breaching stuff paid for by you know risk management center and all the nice stuff buy stuff paid for by krell actually corel yeah cold regions research lab wanted to use raz but they wanted ice capabilities to do some of their own research work and so they finally put ice into raz and i got steve daly from corel he's retired now he's a little older than me he was instrumental in getting ice in her ass if it wasn't for him it would have never happened so yeah it's pretty cool i think a lot of people just think that you know it's just these mad scientists down at hec coming up with all these ideas on their own and making it happen but it seems like it's much more of a a collaborative thing between a lot of different people all over the country and just users that don't even pay for stuff you know ideas come from users ideas come from us too you know i've had lots of ideas and so is mark and steve and cam and everybody in the restaurants had ideas for cool things but a lot of the ideas have come from users you know and then we just flip the light bulb on oh yeah we can do that yeah yeah yeah what i always tell people when they say hey why doesn't raz include this i'd say hey send them a note if it's easy to put in and it's interesting to them you know they'll put it in if it's a big deal you better be ready to pay for it though yeah that's one thing that you don't know how many times someone's come up to me said hey i sent you a email like five years ago about a new feature and you still haven't put it in i'm like [Laughter] what were you asking for you know you know we get asked to put tons of features in and and hey you got to have funds to support it be it makes sense right right that's the other thing and um and then there's priorities there there's an endless list of features that should be in hc ras that are not there it literally is you know just because time and that's money that's a great transition gary for talking about historically implemented new features who funded it and now we can talk about uh one of the exciting developments in the hcc ras world which is the new release of raz version 6.2. so i'm going to go ahead and share my screen so that everybody watching today uh can see these new features if you haven't already heard some really really cool stuff that came out of this news release so i'm just going to go through these one at a time rapid fire and i'll let you guys chime in if there's anything you want to add or or give some some clarity to why this is this is so important to to hcc rise as a program um so the first is we now have the ability and this is in beta form it looks like to have horizontal variations in manning's values across 2d cell faces correct me if i'm wrong gary but previously the manning's m value at a cell face was determined what the meaning's value was determined based on what the manny's value was at the center of that cell phase is that correct yeah that's right okay and now we have the ability to have a composite manning zen value based on uh the stationing of that cell face and kind of what mannings and roughness values line up along that face so this seems like a pretty cool development yeah absolutely i mean we had always planned to do this right from the start but for the first like beta version we thought okay well let's just pick up a single end value per face because the 2d model most of the time people are going to have smaller cells and it won't be that bad of an approximation anyway but after raz was released people started using larger and larger cells and it was obvious that you know in lots of places especially over bank areas people are using larger cells and it's obvious that it should have had you know multiple end values and so this was something we always plan to do it should have been done a while ago so i got to give credit to mark jensen on this because uh mark after i left decided he was going to push this through and good for him because it's a extremely needed capability it's been a long time coming yeah this is great i really like this feature but i was really hoping to see vertical variation in that phase that's the one i really want to have that's already in the design works too and it's going to be vertical with respect to depth per face um but again this was the first step and then vertical duration the depth would be the next step yeah because the way the way you had the way you brought that up gary is interesting and i have to ask you're obviously not working for hec any longer you're working in private consulting do you still keep tabs on kind of what's coming down the pipe and have conversations with some of the folks there or you just like the rest of us just kind of getting the release notes as they come out no i call them all the time because the reality is you know i'm now a user right and so you know and like when six one came out there were quite a few bugs in it and i was like what the heck you know so i called what's going on here i got and i started just sending them bugs and but the other thing is i i you know i worked with those guys i hired every single one of them okay and i worked with them for you know mark and steve since conception in in 91 um and so i i go over there frequently to have lunch like i was there today i had lunch with the razz team today that's so cool so um and i am going to be actually working with them under contract um to do some things on raz so i am going to be uh involved going forward a little bit but i have been involved just you know by calling and contacting and talking about bugs that i found and and this and that and they contacted me a few times with questions and and things too so it's been really good it's been kind of a two-way street awesome nice the next feature here is um all about render mode improvements so chris i'll kind of let you drive the conversation on this one um in terms of the improvement that was made to to how results are displayed in raz mapper well i mean yeah gary can talk more than i can on this but you know i my impression is it's probably a difficult thing taking the computed horizontal water surfaces in each cell and somehow making some intelligent decision in the code to translate that to some sloping water surface and that's great probably are easy to do maybe on 90 percent of the terrain but you always have the the terrain that screws it up right makes it difficult so i'm guessing they've just expanded the ability to provide a better surface or more realistic surface right gary yeah um this this change though was that a specific problem that has to do with rain on grid so the the default mapping mode in raz is called hybrid but it's really sloping plus a little couple off shoots for when you're near a levee that it doesn't slope over a levee it goes flat towards the levee and so forth but when you have rain on grid let's say you have a real you know and you have a kind of a v and you have some over bank that goes way up when you have rain on grid all of a sudden you have a water surface in every cell but it's shallow so every cell says oh there's water i can connect it yeah you know and you get these fill-ins of a real that are way more volume than is really there just because the sloping rendering mode is connecting water that shouldn't be connecting because of the sleep terrain slopes okay and so we were looking at various ways to solve this problem and alex kennedy i got to give him the credit mark jensen um came up with some ideas on how to improve that and that's what this is so very cool yeah next uh the next feature here is uh it looks like the ability for users to download usgs terrain data directly through raz map as opposed to having to find terrain data sources independently save those to the project folder where your raz project is at and then bring those in you now can just kind of on the on the fly bring those in i imagine that in most cases it's not super super detailed terrain data but at least it's something and it's very easy to bring in well it is going out to the usgs site where they have what they call their best available data okay now sometimes that may be good data because it's newer data but sometimes it might be a 10 meter dm right and it is but a lot of times you want that 10 meter dm anyway to supplement where you don't have really detailed data so you want that as a kind of a base anyway yep and so this is a great feature and this is um cameron and anton uh can't put this together for six two so cameron ackerman and anton rudder siren and i think it's great i've used it a couple times already works really well and it's going to expand this is like the start right as other online data sources can be programmed against they'll they'll start adding them yeah now it's got to be kind of like publicly available sources to you know um the other thing i hope they do is they add just kind of a text file url type of approach so they'd have to make this more generic probably to do that but if you had your own data server that you could put in a url and do that we did that with some other information but that might be a little bit harder but that would be a cool way too because then you could add your own servers whatever they might be into your local ras yeah yeah yeah one that comes to mind just being in oregon here is we have the dogami lidar library uh of lidar datasets and they'd be cool just to be able to copy paste that url here and then be able to access all those those files that'd be sweet yeah now this editor is doing a lot more than just downloading the data you know it's bringing it in to and allowing you to create a train model out of it so there's a little bit more going on than just you know downloading it it's making razzmapper aware of it and etc etc um so it's pretty cool i'm gonna use that i'm looking forward to trying that out that sounds really cool awesome next here it's all about sediment transport uh so we have the sediment rating curve analysis tool anything special to add here to this feature um this is stanford gibson's work and stanford's always wanted to add this sediment rating curve and he was really jazzed to get this in he's wanted it in raz for quite a while and there's a new computer science kind of team zach that worked on this so i want to give zach credit um and you know this is something you need in cinema transport work to do model calibration right and um and look and you know bringing in load data et cetera so yeah awesome then we've got a couple other go ahead you guys who haven't seen stan's get sam gibson's videos uh he he does a great job putting out videos of all these new sediment transport features so give those a watch very cool i like that thanks chris for for reminding folks that those are available and then yeah the last couple of ones here are also awesome speed improvements which are always really great for users and raz i'm sure everyone will be appreciative of that we have new floodplain deposition methods again that's going to be related to 1d sediment transport functions multiple maps so this is kind of the idea of instead of generating a single stored map at a time you can now have the ability to to generate multiple at a time which is great then we get into one that chris i know you're kind of excited about and that is uh lava lava flows being available in raz you want to talk more about that well this this one just like blindsided me i had no idea this was even a thing that was under consideration all of a sudden it's oh raz can do lava flows and i started thinking well okay i better start looking for work that involves modeling lava flows i'm going to be traveling to hawaii and the canary islands a lot coming up here but gary were you involved in the start of this or is this something brand new i had nothing to do with this at all i imagine this is stan gibson and alex sanchez that's what i those two are probably the two that put this in and i'm sure that somebody approached htc that hey you can do mud flows but what about lava flows we have all these lava flows you know and that we want to model a lot of those going through you know towns and stuff like that that would occasionally see happen well in reading this though and gary i don't know you may know the answer better but um my impression is from reading about it is it it doesn't simulate the actual solidifying of the lava so it's not going to change the terrain at all it's just basically giving it a different viscosity and allowing it to cool and change viscosity throughout the simulation is that right yeah okay yeah but i don't i don't know all the details either because i was not involved in it at all that was after i left i was gonna say it is an alpha version so it's brand new and probably buggy i guess huh gary i think they thought it was cool so they probably said hey let's put it in 62 release anyway we'll just call it alpha if it's means it wasn't even tested enough to be called beta yeah use it you use it your own uh apparel basically right right in other words it's probably fun to try out but uh you better wait for the next version maybe yeah maybe an external beta i'll just i'll just say what everyone else is thinking and that is how did we get lava before we got depth depth averaged manny's roughness that's what i want to know priorities man what sounds more sexy if you're a razz user i think depth averaged uh manny's roughness that's pretty awesome so you're showing the dirt in you right there ben it's not sexy not at all uh yeah the last couple here uh unsteady rip rap analysis so we now have the ability to do a user rip wrap calculator um using unsteady flow analysis which is really really cool i know that there's been a couple projects that i've been involved in where we wanted to use the rip wrap calculator but we were doing a steady flow analysis and so now having to create a separate plan to do that is pretty sweet 2d bridge stability looks like some just improvements to the overall computations on how that's done i'm sure that's one of those features given how new 2d bridges is to htc rise that probably every version from now until 7.0 there's going to be little tinkers with how that's made and improvements to that overall computational regime yeah for sure yeah and then uh the last one here is kind of funny the north arrow is now available in raz mapper to be added and it feels like this was one maybe that over time you know there was uh tens of people that were emailing gary what about a north what about a north dakota and then eventually somebody's like okay fine we have to eventually do that because literally hundreds of people over time said why can't you put a north arrow over on the map they finally did fine we'll put it in there stop bugging us right exactly exactly so yeah overall bunch of cool features i hope everybody's excited to jump in and try some of those out um besides you know the new 6.2 release which is obviously got a bunch of cool stuff and i know gary we've touched on this back in episodes 12 and 13 a little bit but anything in particular that you're really excited about that you know might be coming down the shoot uh over the next couple years oh yeah there's there's lots of stuff um so one thing i in my new job i've been doing a lot of work in a hurricane modeling so we're working down louisiana louisiana water initiative project and you know razz we added wind forces but um we're still kind of using combined adserk for coastal and razz for kind of the boundary to take ad strict boundary conditions and then come inland with that but i think hec needs to add uh pressure force differential for hurricanes into raz and that could be an easy thing to do and i think they actually are going to work on that and then probably later on well down the road you know look at pattern forces due to due to waves okay and if that were to happen then you could you know razz could pretty much be the coastal model as well as the inland model um but htc's you know even before i left hc's been working on a new gpu compute as an option so uh a cpu compute and i imagine the first version of that will just be 2d cells only you won't be able to have hydraulic structures and then they'll probably add structures later but a gpu compute could could dramatically speed up the computations you know 10 to 100 times for these larger models where you're doing rain on grid modeling especially um to get new computers for that gary well there'll be a certain type of graphics card compatibility that'll be required but all kinds of graphics cards should work and then but if you want to run faster you're going to say well my current graphics card only has like 60 64 gpu cores you know and i'm going to get one that has 256 or 512 then i can run a really big data set fast you know thing so just be it'll depend on your use case right um another thing that's been going on before even before i left was floodways in 2d so the agency's working on tools to try to assist in making doing floodways easier in 2d modeling which you know it's a difficult subject but it's happening at some point they're probably going to work on a simplified groundwater model so you can have groundwater service water interaction and then right now the infiltration capabilities in raz are just um event-based where they subtract the infiltration from the rainfall and you get rainfall access and there's no further infiltration after that and so i imagine a future version they'll have continuous you know infiltration based on the water on the land surface but if you do that you got to have a groundwater component yeah in other words otherwise you just never get any water back out okay into the river so that's probably something that will happen in the future barry i know a lot of people are wondering this and i've gotten this question a lot but uh is razz ever going to have a 3d option [Music] sure why wouldn't it right well yeah i mean who would have thought 2d like 20 years ago right well who would have thought 1d on steady flow when we started even yeah some people as was just a replacement for hc2 right and it was pretty steady but that was that was never the plan the plan was to river analysis system whatever that means you know um and obviously it has limitations uh given where it's at right now but let's face it modeling is you know people are moving more and more to 2d modeling right now right it's because there's lots of software out there that can do 2d modeling efficiently now and then people are moving to 3d modeling more and more too but for right now still certain types of problems and limitations of size and dimensions and and durations but as computers get faster people are going to want to do more and more 3d modeling so yeah raz will have a 3d component someday if it doesn't i'd be disappointed so is it is it more like carries more like one year out or two years we're like five years out or more yeah i was yeah i tell people don't hold your breath but hey you never know it could come out it's good to hear that that's something that's a possibility in the future that's great i'm not saying they started on that because they haven't i'm not saying they're fun to do it because they have they're not i'm just saying that's natural evolution right yep and of course the funding thing too right right speaking of future developments that are very exciting uh full momentum has a really cool uh announcement to make and it's about a new feature that our video podcast here has and that is we've been able to move uh our recordings not only onto youtube where they are and you can watch the video components to them but we actually have uh podcast versions now on both apple podcasts as well as spotify this was a request that we have from a couple of listeners that you know liked to watch the videos but they also like to you know listen to them on their drive to work or while they're doing some yard work which man alive if you're doing yard work and listening to talk about raz bless your heart um but folks folks wanted it and uh and we listened to we listened to those requests and so now you guys have that that option to listen to us uh on the go so if you're interested please go subscribe rate and review us uh if you can uh we obviously enjoy all the uh all the followers that we can get on any platform that works for for everybody that's great to say your your segways are on point today and you have been like you've got that down but you guys keep you guys keep setting me up and i'm just taking what i get so uh very cool we're like i said we're super excited to have gary on not just to talk about new features and other things but we're going to get into the weeds on some technical talk but before we do that we're going to get into some raz trivia so for those of you guys who are wondering we're going to do some razz trivia here in a second where we will each ask each other at least one question um i will likely be stumped by chris and gary but nevertheless i will still participate uh but before we do that let me give a quick read for our sponsor today so we are thankful for our sponsor kleinschmidt associates uh who is known throughout the industry as a firm that provides practical solutions to complex problems affecting energy water in the environment environment you can learn more at kleinschmidtgroup.com so shout outs to kleinschmidt as always for sponsoring today's episode but with that let's get into hec razz trivia uh i will go ahead and start us off so that i can't be embarrassed to to start off this segment so what are the ground rules what are the ground rules here so the ground rules for the for the trivia game today i'm glad you asked chris so uh each of us will ask one question at a time and we'll go around um we'll give each other 10 seconds to think about the answer and then uh whoever asked the question will will ask each individual what the what their answer is it'll be honesty policy in terms of if you what your answer is and then uh the ant the person asking the question will reveal the correct answer there will be no scoring because i do not want my record so uh so this first question for you guys probably will be a softball but i'm interested if you can get the exact exact timeline here so and i actually this is from an online uh resource that i'm not 100 sure is correct but i hope it is so uh htc raz added 2d modeling with the version release of hcc raz 5.0 correct yes data okay what data what no this is version 5.0 so what year and what month was htc raz 5.0 released before you answer are we talking about 5.0 official release or 5.0 pages okay did you say year and month that's what i said oh my gosh that's not a big deal i mean i know i know the range but i'm not going to get the month that's okay you get partial credit okay let me think about this gary looks like you know i know the year of course okay uh i think i know the month but i'm not i'm actually not certain i don't remember the exact month to be honest but i think i know it okay i'll have a guess i'm going to guess what chris is thinking gary why don't you give us your guess so the year was 2015 for sure but i think the month was march but i'm really not sure that at all okay so my guess was going to be 2014 november but i i you know between the beta and the full release it's a little hazy for me so um 2014 november that's my guess you guys were both close gary you were closest so the correct answer is february 2015. are you sure wait a minute are you sure the manual doesn't say february but we didn't actually release it until march because that has happened there's a there's some gray there's some gray here for sure for sure yeah that's a good one why don't you go next all right so i've got a bit of a nerdy one all right um but i think ben if you think if you think deeply on this you may be able to get this and gary i don't know you may know it you may not when you say think deeply you mean like think into google and ask the question or [Laughter] exactly that's what i did all right so um as you both know that in unsteady flow modeling both 1d and 2d raz has to use matrix solvers right to solve the the series of equations that are developed from the conservation of momentum and conservation of mass right so there are many options the default in 1d is the bar cow right or the skyline matrix right and the default in 2d is the pardeso which everybody calls it paradiso but it's the pardeso so my question to you guys is what does hardiso stand for and i'll give you a hint it's three words it's an acronym what does pardeso stand for p-a-r-d-i-s-o ben looks like he's about to uh have a meltdown here i'm going to say pair of two solvers so so i just broke it up into three you're actually very close ben very close gary do you have a guess i do not know the answer i'm gonna be honest i should know the answer to that but i do not know the answer all right i've stumped you both awesome direct solver so you got the solver and i think what did you say you said pair of somethings all right or partial or something anyway you were pretty close ben so well done parallel direct solver yeah all right gary let's hear yours okay i got a uh an interesting one for the first one what is the only feature ever added to hcc raz that is site-specific and not generic meaning it can only be used in one location on the planet oh oh oh i think i've run into this before but i'm spacing which feature what is the only feature ever added into hcc raz that is site specific not generalized and can only be used at one location in the entire planet is it still in there can i ask you that is it still in the current version it's in the computational engine but it's no longer accessible from the it was in interface for a long lot of years so there are two that when you started asking that that immediately came to mind but then i realized they're generalized but go ahead man so is it i'm gonna guess it's something to do with the eagle creek area that you grew up in you mean bald eagle creek body okay yeah yeah no it's not bald eagle creek though good good try though chris all right so these are going to be wrong but these these were the first ones that came to mind because i know that they're generalized but i immediately thought of the navigation dams feature that was put in i think it was originally like for the mississippi river but obviously applicable to others and then i thought well maybe um that uh danbridge repair option because i know that was put in for uh new orleans district right but as far as something specific i feel like you're going to say and i'm going to go oh yeah i remember that because i i do sort of remember something with a name in it but uh let's reveal it i'm stumped well you were close because it was for the new orleans district oh it is it does have something to do with the mystery oh wait wait no i do i know what it is toulette man it's the old river diversion the old control structure that's right yes that is used to split flow between the mississippi and chapalaia to maintain a c30 flow split interesting that's right gary the quick movement wrong when you download some of the initial razz piles when you first download rats from the internet one of the things you can get is the eagle creek terrain data set correct bald eagle creek political creek quality train data yeah yes so would that be would that be considered site specific then but that's not a feature in iran's software that's just the data set okay i'm not a computational feature i should have clarified a little more computational feature yeah that was good i had no chance so i'm gonna pass on my second one chris because i was you know busy running the the podcast while you were looking this up so i'll let you go for your second one here so okay so my next one is is i mean it's sort of razz related it's more razz team related okay and uh gary should know this and then you might have a guess on it in fact gary this should be in your calendar somewhere but um what was the month and year that i started at heck [Laughter] oh come on why should i know this let's see well you were just talking about this on the drive over to eastern oregon so yeah i get my guess is gonna be um 2002 i don't know the month okay i'm going to say 2 000 oh hang on a second i should i have a way of figuring this out because you know i have a way of figuring this out because you are the drummer in our band yeah and the last album released so it was probably um 2001 or 2002 2001 i think i'm gonna go 2001 you started i don't have a guess on a month spring you know so let's say june june two oh my gosh gary you're so close you're right about spring it's may 2001 okay may 2001 nice job and ben i'm impressed because this was back when you were in uh maybe kindergarten or something uh i think it was in yeah yeah third third grade sounds right okay long before you uh you started your hecara's days but um still anyway yeah what do you have a second one gary i have a second and i have a third even okay so the second question is how many official releases of htc raz have there been not including beta versions and okay define official what is an official release is that 1.0 was an official release well is like 507 an individual official release absolutely okay oh my gosh okay i'm going to just think with beta versions though um let's see i'm gonna say i like that number ben um and i hadn't thought of a number first so i'll be honest so i'm going to use your number and i'm going to play you're going to probably play prices right here and i'm going to say 20 what did you say 22 i don't know what i said i don't remember 22. oh i'm going gonna say 23. guess what i should really give this to ben but it was 23. [Laughter] funny prices well done man i i have no idea what i would have picked if you yeah you came up with 22 really quickly i thought you guys were going to start thinking oh okay 1.01 you know counting them up but you don't even count i just thought i was thinking you know there's there's we're up to six we're up to six point two now so six main versions and then maybe about three releases per and then a couple extra yeah that five version you know five five oh one five like two all the way up to 507 there was a lot of fives yeah yeah hey ben have you been have you ever seen a screenshot of the very first release of heck rise the main window it's kind of fun sounds terrifying though i'd probably be scared it's pretty funny but it's very similar it doesn't look that different i mean it's funny well it's like really short it's really narrow and then it had to expand out with all the features right hey we don't we don't we don't sideshame here on mark jensen and i painstakingly drew all those icons and you know and all that stuff on that first province all right last question okay and this is kind of similar to what chris asked but when was the date exact date month day and year of the very first version of raz release not beta version 1.0 and why was it on that date oh version 1.0 um [Music] and there's a there's a reason it was on that date yep probably gary's birthday no it wasn't my birthday oh let me think um i'm gonna say uh june 1st 1990 i am going to say yeah i can't think of what is i mean even if i thought of what the the occasion was i wouldn't know what the date was but i was thinking maybe it was like on the uh anniversary of the founding of hec or something like that um but i don't know what that date is so i'm going to take a wild guess i'm i think i'm close on the year uh as far as month and day i have no idea so i will say january 8th 1998 okay uh well the first beta version went out in 93. the second rated version went out number four the first version went out in 95. okay and it was july 8th of 1995. and what's going on the right day anyway yeah you had the right day what's the name about that that was my 10-year anniversary of working at hcc wow so i'm very close to that date so i kind of forced it to be the first version to go out to live so the first version the first the first version was released on your 10th anniversary and your last day of work at hcc was a racist xl so yes that's kind of cool a couple of bookends right there huh there you go they should just rename the software hec gary instead of hcc rads cool awesome well that was really fun guys yeah that was cool that's very interesting i didn't realize it took uh you had two beta releases before the full release of version one came out yeah yeah that was just steady flow right yeah the reason it took so long is that i knew right away that if we weren't able to reproduce pretty much all the features of hc2 people would not use it and because that's just life right they'd say well that's nice but it doesn't do this it doesn't do that so we really held off until we had not every but like 98 of the features that htc 2 had before we released it plus obviously a graphical user interface and graphical output and all kinds of other computational features that were newer so yeah great that's really cool all right well without further ado let's tie into the technical topic for today which is going to be about utilizing hec raz for title modeling this was actually a question that we've gotten from a few different users it's also one that when chris and i first got we're like don't know a ton about it so chris spent some time kind of diving into the model experimenting with some things and he's going to show us some examples of of some things to consider when you're building a title model and then gary's going to look at what chris did and tell us what we did wrong so [Laughter] chris i'll turn it over to you yeah so i've you know i've been doing title modeling for a long time and i've had many conversations with you gary about you know things best practices things to look out for you know certain settings that are important things like that and so i kind of had this baseline background knowledge but i never really had an opportunity to explore these different techniques and different ways of of setting up your computation so that you get the best out of your title modeling so i thought i'd put together just a real simple flume type of model and i'll show you what it looks like right here and zoomed out it just looks like a straight line but this is actually my 2d version of it and if i zoom in like down here on the downstream end the tide side you can see it's actually a 2d model here i'll get even closer so it's made up of a lot of cells but it's a very long and narrow prismatic channel basically a trapezoid that just goes for i think 100 kilometers upstream so very overkill but i wanted to make sure i could apply this to a lot of different types of applications there's also a 1d version of this because i looked at both 2d and 1d modeling and so uh the 1d obviously is made up of cross sections instead of the 2d mesh so i'll show you what that looks like here okay so here's my 1d model and in fact when i wanted to build the 2d model i use the 1d cross sections to generate a terrain which the terrain you can see behind the cross sections here so this terrain is actually a surface developed from these cross sections and then i use that surface to extract terrain onto my 2d mesh for the 2d version of this so got this very simple thing all we've got on the downstream end is a simple idealized tide cycle so if i open up the unsteady flow editor you can see we've got a stage hydrograph on the downstream side and um i know this because we just asked gary this a little earlier but this is the best way to simulate a tide is with a stage hydrograph so some people try to force a flow hydrograph um it's a little bit tricky plus how do you determine what the flow is for one uh maybe you have a gauge but it's it's better to put a tight cycle in with stages so here's my stage hydrograph you can see it's let me make this a little smaller so you can see the whole thing it's um a diurnal type meaning you have two peaks and two troughs in one day and i know gary i know you're gonna say this but the the high the two highs are not the same every day and the two lows are not the same every day it changes so i understand that this is a very simplified version of a tide cycle so i just put that in and then base flow on the upstream end just a base level 10 cubic meters per second okay now my question gary this is what i want to get from you in this is what are some of the important computation settings what are important equations to use what are some of the things you want to think about when doing a tie that maybe are not so important in just regular river modeling well you know tide modeling is true wave modeling and yeah what when you do kind of this kind of wave modeling wave propagation you really got to pay close attention to cross-sectional spacing or cell spacing and time step first of all so you need a small enough cross-section spacing or cell size that you're able to pick up the dynamics of the wave rising and falling uh accurately and you're not uh numerically attenuating it because your distances are too far apart a then b your time step okay if you're using too large of time step then you're going to attenuate the wave as it moves in just because your time steps too large so those are first and foremost then after that there's a parameter called theta in ras that whenever you do title modeling you need to evaluate theta and ras theta is set to 1 by default because a theta of 1 gives the the solution the best chance of solving stably right there's that theta implicit waiting factor and we set one because reality is in real world modeling you know away from tidal modeling most rainfall runoff types of floods theta 1 gives you not only the most stable answer but gives you a very accurate answer and and making it less than one really isn't necessary for most situations where it does become necessary is when you have highly dynamic situations whether it's like a dam break flood wave or a tidal boundary condition and or flash flood or you're opening and closing a gate really fast so where you have really rapid changes in flow and staging velocity over short distances and times and so this theta factor changes how it weights the derivatives the spatial derivatives which we could get into and in theory the actual theory is that a theta of 0.5 would give you the most accurate answer where you're waiting yes let's talk about this diagram let's first let's explain what this diagram is so this is what's called the space-time diagram and on the bottom is going horizontal as space and let's think of j as a location j is a cross section and j plus one is another and then vertically it's just time so let's think of n as the current uh or the previous times that's already solved and n plus one is the time step we're solving for so down at j and j plus one time n let's say we already know the water surface the velocity everything okay and now we're moving forward in time to n plus one and we're trying to solve for the water surface velocity everything at the n plus one time line at all space well by default raz is going to compute derivatives for change in velocity change in pressure forces spatially and change in friction forces spatially based only on the currently solved timeline n plus one so it has to make a trial okay and it gets those velocities and then it can get the change in velocity um and change in pressure forces and change in friction and put that into the momentum equation and solve it just to clarify gary you you said n plus one you meant the current timeline is n right well the the solved time line is n now we're going to solve for n plus one so n plus one is the unknown time line n is the known we already have an answer okay yeah so the theory about theta though this is what's called a box scheme and the box scheme if you go and look it up and look at the theoretical analysis and we're not going to get into all the details but in the in the theory of it if you actually weighted the derivatives that n plus 1 and n equally in other words theta was 0.5 that's the most accurate answer for tracking how something changes with respect to time over the distance if you think about it think about velocity changing with respect to space and time knowing the velocity at j n and then the velocity you're trying to get to how that velocity moved through that space over time to n plus one it makes sense that weighting those derivatives equally from the new timeline versus the previous will give you kind of the most accurate representation of what happened over the time step but unfortunately numerically it turns out using theta 0.5 while it's the most accurate it's also the most unstable okay and unless you use really strict time steps it has instability problems whereas theta of one you're using derivatives based on the current timeline only to see what happened over the time step isn't as accurate but it's way more stable and so that's why in raz we default to one but for certain conditions we talk about that you should really investigate moving theta down towards point five in raz we don't let theta go to point five because we know that's it creates some instabilities and we've done lots of theoretical testing on this and theta point six basically gives you the same answer as theta point five but it's just a measure more stable so in raz we only allowed you to go uh to between point six and one for theta so you're in this direction right here right gary yeah exactly theta theta equal to one would be it's using the n plus one time line to develop the derivatives the spatial derivative so you just need to take space right you'll be taking the velocity at j n plus one and the velocity at j plus one n plus one and and you'd be you know looking at that change in velocity over that distance same thing with the pressure forces at that point in time and same thing with friction losses at that point in time so here's a question for you gary why or how in the world are you going to use n plus 1 if that's in the future time where you haven't computed it yet well it it makes you know a guess at a water surface based on the previous derivatives okay so it's kind of got a slope projection already from the previous timeline and it uses that as a first estimate of what what might change between the previous time step and this new one and that's the first trial sometimes that works out great and it solves it within the tolerance says oh good enough i'm moving on to the next time step a lot of times it's you know maybe not good enough and it and it uh says no uh that slope projection wasn't accurate enough so things are changing maybe the flow rate has a certain change or or something like that over the time and so it has to you know update that those guesses and try again okay and this is all thrown into the matrix solution right yeah because it's solving all of space in each time step so the questions are written for every location and there's dependencies upon you know cross section next to the other cross sections for calculating these derivatives um and then it's thrown into a matrix and then upstream you obviously have a known flow coming in at the new time step and then downstream where you generally have a boundary condition and in the case of title that would be a stage right you already know the stage at the new time step because it's your boundary condition yeah so there's iterations upon iterations and there's matrices being developed and inverted all throughout so all of you razz users out there next time you complain about how slow your models take and keep in mind what's going on in the background here yeah there's a whole lot of numerics going on just to do this even in wendy then let alone 2d yeah that's that's great gary i appreciate the explanation of that box scheme because it's and it can be applied to 2d modeling as well right yeah same concept you just have two spatial directions instead of one right yep so it's a cube instead of a box i suppose yeah yeah oh very cool um okay well why don't we uh get into the model that i built here and you can tell me what i did wrong or what what could be done better um but basically what i did is i set up several different plans and i wanted to compare different thetas because theta is important in title modeling as gary just mentioned i also wanted to look at things like well what about 1d versus 2d how's that going to change the results and why would they be different and then diffusion wave versus elm uh which is the the more commonly used full momentum version of the uh symphony equations that are built in and so wanted to take a look so let's uh gary why don't we look at how about a 1d model first and i'm going to just compare theta equals to 1 versus theta equal to 0.6 just to see what the differences are and what i've done here is i've in this model i've drawn some profile lines so we can extract some data off of it and i'm going to take a look at a transect that's halfway up the length of this model that's why it's called transect 50. it's halfway up and just look at the flow hydrograph differences between those two so pull that up here and you can see right away there's some differences between theta equals one oh actually i've got some other plans on here hang on let me uncheck those first 2d plans okay let's go back to that okay so with theta equals one and one d not zoomed out not a whole lot of difference but if i were to zoom in and tell me what you think about this gary let's zoom in pretty close here there we go now we can see the differences and it looks like way out on the fourth i think the fourth cycle tide cycle it's about maybe one tenth of a foot different or a meter sorry one tenth of a meter difference so 10 centimeters maybe not a huge deal but significant enough yeah the other thing about this is um i'm betting that you picked a really good cross-section spacing already and have a really small time step yep so you've already taken out of the equation diffusion due to do-do excuse me is that a word doo-doo [Laughter] let's not get too technical gary yeah just in case some people are following yeah um so you've kind of taken out the equation already any kind of attenuation or numerical diffusion due to the space distance being too long for the wave being modeled and the time step and when you do that differences because of theta becomes smaller too so in other words if we had used a larger cross-section spacing and a larger time step we might have had some numerical fusion we didn't even know about because we had too large of a cross-section spacing and time step right yeah but then we just did the theta test only we would probably see even bigger differences because of theta also at this point so shorter or closer space cross sections and closer spaced cell centers i is a good deal or a good good thing to work towards here but um along with that you've got to have a correct time step too meet the current member condition right right yeah so i know everyone's thinking well how close do i need to make my cross-sections or my cell centers and and um i'll answer that and let you gary after that but i mean you really don't know for sure until you test out some different spacings and see how the results start to converge is that about right gary yeah that's really the only way to do it and what's funny is back when computers were really slow engineers used to do that on a regular basis they used to test cross-section spacing and time step and it was always part of their study to do that to make sure that they weren't getting numerical diffusion okay it seems now that computers are so much faster kind of people are kind of forgetting that part of the fact do i have enough cross sections are they spaced close enough together is my time step small enough and you're exactly right the only way to check that is to do a couple of trials so let's say you have 500 foot cross-section spacing well you sure it's actually run a run with 250 foot spacing and an appropriate time step to go along with it for a current number based on that smaller spacing and then compare that to your 500 foot larger time step run and if there's a significant difference that tells you okay your 500 foot wasn't good enough and then you don't know if your 250 foot's good enough unless you cut that in half maybe again okay and go down to 125 foot but it should converge as you get to shorter spacing and smaller time steps the numerical diffusion to the spacing cross-section spacing or cell size and the time step should be diminished towards zero and that's called spatial space and time convergence okay and that's when you really know okay i've got a reasonable spacing and a reasonable time step i can move forward from here and in theory you should see your theta differences decrease along with that decrease in computational load spacing right yeah because as you're reducing the time step the change in the derivatives is less over the the both and produce reducing space the change in derivatives is getting smaller right yeah and weighing them between timelines the timelines are closer together that's smaller so theta should become less of a dependency it might still be a dependency depending upon the problem okay and how dynamic the wave is like if you have a dam break that goes from nothing to a million cfs in five minutes well data is probably still going to be a factor okay and you might want to mess with that and check it all right very cool very cool let me move on to another example let's compare 1d to 2d and i'm just going to go with the theta equal 0.6 because it's kind of where you want to be for for a title model and i'm going to go to a 2d but i'm going to use diffusion wave first and i've got this set to 10 seconds for my diffusion wave model and it's all based on trying to get a corrupt number down to one and i know that's tighter than you need for diffusion wave but i wanted to just kind of take that out of the equation altogether so we've got 1d theta equals 0.6 compared to 2d diffusion wave theta equals 0.6 and let's see what the difference is here now we got a big difference so this is telling me that there's something different obviously between the 1d and 2d equations besides just the additional dimension because this is a very one-dimensional model right it's just a trapezoidal flume so what why would there be such a difference between these two gary well in the diffusion wave equations we don't include the the time derivatives is that that's you know the change in velocity with respect to time okay uh we're only including um an or change in velocity respective distance we're only including the pressure differential and gravity force uh and and uh friction forces and diffusion wave those are the forces so you don't have those velocity terms in there they're dropped out okay so it's you know it's quite a bit simplified and you know if you're working in a river system where the slope is fairly steep well then gravity and friction are really controlling almost everything and that's where our diffusion method will shine is in you know steeper slopes okay at least medium slopes but once you get into flatter slopes and you have you know more kind of like wave propagation and what has happening downstream affects upstream and especially if you get into title you know title coming in is all about velocity right as far as how much it's going to propagate up and propagate back okay so if you don't have those velocity terms and derivatives then that alone is why you're seeing this difference and really more so than anything related to time step or theta uh it's it's the fact that you don't have those velocity terms yeah i've always heard that you know when you get into really flat systems and you know with backwater being a dominant part of it and you know what's more back watery than a tidal type model right is those inertial terms on the left-hand side of the momentum equation are extremely important and diffusion wave gets rid of those so that's why the diffusion wave is not a good option here yeah and this is and this also is for you know just a simple trapezoidal channel okay i think in natural channels it you should probably even show more difference yeah so let's then look at a full momentum solution so i'm going to just switch off the diffusion wave and i'm going to turn on the elm theta equal 0.6 still and let's see what we get here okay it's better right there's still some difference but it is better yeah i mean the the timing and the shape is closer there is a little bit differences in the peaks and the troughs okay and um the elm is the elarian lagrangian method and that was our first 2d method we came out with um and it uses what's called tracking to track upstream to find the change in velocity so you have a face in a cell and it tracks upstream over the to find where and spatially did this velocity start and how far did it travel over the time step and what was the velocity at the beginning okay and that tracking unfortunately you know it has to do things like track back bases but also has to do like estimate velocities in the middle of cells and so there's some numerical averaging happening and because of that kind of stuff you can get some numerical diffusion of that velocity term that inertia term okay um in most real world models rainfall and off and so forth you're not going to see much especially as you make the cell size smaller and the time step smaller but where it does start to show up is kind of these difficult problems that are highly velocity driven okay um and you'd see it also at like for dam brake flood waves you might see a little bit real tight contractions and expansions where you have rapid changes in velocity versus distance um then the emf is the uh just the alarian method that method was we added um in 60 because we you know we realized that this current method that was in the original raz 5.0 had some locations that you know it had some numerical diffusion um and and that method was really developed to be a fast and b stable those were kind of like two of the driving forces it's not that actually was important of course it is but we really want it fast and stable especially for wetting and drying which does really well okay for 2d et cetera but then we decided to add a third equation set which is the em method and there the whole goal is you know we want momentum conservation period and it's an explicit method and you have to use smaller time steps in general than the elm and it evaluates the velocities only locally around the face halfway upstream and halfway downstream and you can read about that in the hydraulic reference manual so it doesn't have to track back in space and then and interpolate and estimate velocities not estimate but interpolate velocities so it does a much better job at conserving momentum so i bet if you go to turn on the em answer it should be better than the elm for this problem yeah let's do that and before i do the the l in elm is for lagrangian and that's the tracking part of it right right the lagrangians yeah the lagrangian part of the elm and then taking that out it gives you the em and um let's see what we get there so i've done the same thing using the em we'll go to theta equal to 6.6 by the way i had to reduce the time step to make it work because gary you mentioned it's a lot more strict with current numbers and and not as stable as elm so i had to get the time step down a little bit to make this work but uh let's see what we get here okay oh still a little bit of difference between 1d and 2d well the 2d is higher in this case oh did it actually dip below oh yeah i didn't even catch that okay so let's turn on let me turn turn the other one on turn all three of them on yeah yeah okay i may have flipped that in my brain well let's make sure we got that right okay so 2d um em is the dark blue line well this this is not helping i can't see where the green line is so let me zoom in really close here and see if we can pick that up not showing it for some reason it might be on could it be i think it's under here i think the elm and the em are the same looks like it's the same no okay there we go so this is showing the difference between eem and elm the em is higher a little bit and then the 1d is you know down here a few meters below that so and this so in this case uh yeah do the full plot okay so this is upstream so in actuality the 1d is showing more attenuation than the 2d solutions yes it is yeah and but it might also get into i mean is your cell size the same as the 1d cross section spacing no that's right your cell size is a lot smaller right yeah yes so exactly and then you have to use smaller time steps also so here it's kind of a case of really it's because the 1d cross sections are much farther apart and using a larger time step you could probably get a closer one the answer if you went down to a much smaller cross-section spacing and used you know a 10 second time step you'd get closer to the 2d answer then yeah that's that's really interesting and i think it just really solidifies the point here gary that especially in a title zone of course i'm doing a very idealistic flume model which you would never see in real life out here which probably had some additional problems you know as far as answers being different but um you got to look at these things right you got to test things out you can't just build your model and take the results for granted that they're correct and then what would you say gary would be the most important things when you're thinking about doing up a highly based model in an estuary something to think about well first of all is you've got to have your boundary condition if it's ocean driven i'd like to put my boundary condition out in the ocean yeah okay and the reason is and i like to have my terrain like ocean level and then coming up because then what what you're going to get is you're not you've got that wave propagating up and down but you're also going to get a convergence of the terrain vertically and you're gonna and then that'll help capture the velocity acceleration that occurs when that wave comes in right yeah and so that'll give you more accurate velocities coming in if you just put the tidal boundary cushion in the middle of the bayer estuary yeah it's still going to have velocity go in and out but it might not be quite as accurate okay so the important thing is where where was the data measured was it measured in the middle of the bay then of course you use it there but if it's if it's an ocean gauge then you know that's out away from the the boundary okay of the the coast and so um i like to add some terrain out of the ocean and then put that boundary condition out in the ocean if it's a ocean type okay so that's important and then as we talked about earlier cross-section spacing or cell size and time step is then the next most significant thing and then checking theta after that but generally i found that if you do any kind of title modeling you should use theta less than one by default i usually set data to 0.8 if i'm doing any title modeling just even to start and then i'll check moving it down to 0.7.6 to see if it makes any difference a lot of times it doesn't when it's sometimes good enough sometimes it does make a difference but definitely every title kind of modeling ever done theta 1 doesn't work good [Music] because what happens is as that wave comes in and moves upstream it it diffuses it and it doesn't show the influence of the tide going as far upstream as it should okay because it's going to you're just showing kind of one location here but somewhere way upstream you know the the worse theta is or the your cross-section spacing is too large your time steps too large the influence of the tide is going to diffuse out quicker and it's not going to show up farther upstream and a lot of these systems around the country like you know the the columbia river i mean at low flow you see a dominant tidal signal right up there portland you know that's a long ways from the ocean okay down here in sacramento you know we're worried a good i don't know what it is i'm gonna say it probably wrong but you know 50 60 miles from the san francisco bay by the ocean by the river and there's a strong title signal at sacramento during low flow oh wow now for really high events that gets washed out right because the flow is so high in the river and dominating the water surface and velocity downstream that the tide may not be that important especially maybe for like a dam break if you're doing a dam break model on a coastal area the dam break flood wave is going to dominate the water surface all the way out to the ocean the tide might not even affect it so it depends on the problem you're modeling also yeah so i kind of see this as you know setting up a model like i did here is a flume it's got very little bass flow coming from the upstream end that i'm really magnifying these differences i think uh correct me if i'm wrong gary but having more base flow coming in that's going to mute the differences as well as having a lot of difference in terrain and variation in terrain and shape of the channel as you go down is going to also have a muting effect on the differences yeah one thing about baseball though it can it can work kind of both ways is deeper water allows ways to travel faster right yeah yeah so if you had really deep water you might get higher velocities interesting so then shallower water you know less effect yeah more more frictional based right well now we saw very little difference between elm and em in this particular case would you recommend em for a model uh a typical estuary type model or we still do an elm for that in your opinion um i think you have to try i i don't want to say a or b because i'll be wrong in some instance right yeah i would say in general that either one will work fine in most title situations but there's going to be some where the em might work better and the only way you're gonna know that is to try them both but the great thing is it's so easy to do unrest create a second plan switch to another solver maybe adjust the time step a little bit run it compare the answers and then you have the answer for your specific problem um but i think most cases elm the original one does quite well and as you can see here the differences were small okay for the 2d one right right yeah well but it could be site specific though so i think the new solver is really important where you have uh sharp contractions and expansions in your river system because the spatial variability of the velocity is dramatic and that's the tracking method it's hard to track where the velocities are coming from and sharp contractions and expansions okay and where they're going and that's where the original method the tracking you know you start to get a little bit of a numerical attenuation of the inertial terms because of that interesting yeah i wasn't aware of that so if what you're saying is if the advective acceleration contraction expansion is dominant that em may do better a better job than elm in a river and again you gotta you gotta test this and know how significant is in some cases it might not be significant right interesting yeah yeah that's great to know um so gary before we wrap this up i just want to throw one thing out and and as you know i was also doing some tsunami modeling in along with this i didn't really want to share any of that today because it's not ready but um the tsunami is basically the same thing we're looking at here only just a very extremely dynamic version of it so um you know you have going from say a stage of zero to maybe 10 meters 15 meters or even more in the span of about a minute or so and watching that progress up upstream do you have any thoughts on what pitfalls you may run into trying to do tsunami modeling and how some of that may affect some of the decisions and how you set up the model um i have to say i've never used raspberry tsunami modeling so i don't consider myself an expert on it at all um but i so i i don't think i have any tremendous insight into tsunami modeling i'm sure there's lots of people out there have much more insight than i do but i would imagine that's definitely an area where you want the terrain out in the ocean okay and setting your boundary condition farther out because when that tsunami comes in you know you want that convergence and that acceleration of velocity uh coming in and again some of the same things we talked about here smaller cell sizes smaller time steps uh et cetera and then test the sensitivity of it yeah that's the key the key really is testing the sensitivity of the space spacing whether it's 1d cross sections are cell size the time step and then in this case theta okay yeah one some of the stuff that i did notice in my preliminary modeling of the tsunamis is even in a profile view so in a profile view for what we're doing here just a regular tie it's pretty boring there's not much to see i mean we could turn it on here um i could just plot a um let's just plot a water surface profile like this and you know i can animate it and it's pretty boring this is the lines going up and down uh pretty flat but with this tsunami you see a very clear and distinct wave and you can watch it propagate up this slope and there's some very clear differences between um diffusion wave and elm and em as you might imagine but also in the different thetas and you even get some pretty interesting artifacts on the front end of that flag wave you'll get some some instabilities kind of a up and down jagged pattern in the wave front whereas the rest of it may be stable but just the front of it is has got a lot of bouncing around and so anyway i hope to present that a little bit later maybe we'll get you on to talk about that too gary well you know the key right there is the the wave is much more dynamic than the title you know you know because you're going from a lower stage to a much higher stage or a very short amount of time chris did you include coriolis in any of these simulations i did not um but if there if you if you were modeling a larger bay or or river confluence with uh with the ocean it might that might be actually one of the few situations where it's worth turning that on right i'd love to hear what gary has to say about that oh uh the answer is yes but i would also qualify that is more so true if you're farther away from the equator right so if you're up in alaska yeah you know corey also might be an important parameter to turn on i understand um yeah i've always been to the opinion gary on the coriolis and you can correct me if i'm wrong here is that in a typical river you're probably not going to see much of an effect and obviously the effect will be more pronounced further away from the equator but it's more large bodies of water where you really see coriolis is is that correct yeah absolutely okay good because that's what i've been telling people yeah and obviously tidal modeling of certain segments of the ocean would be included in that so right absolutely yeah very cool well gary man that has been fantastic i've learned a lot already and um and just our little discussion here and so i'm hoping everybody out there has picked up on some uh some best practices some little theory knowledge as well which is always good to have uh helps you make better decisions and how you set up your model and how you troubleshoot it so thanks gary appreciate that uh thank you for having me on it was fun it's always fun with you guys so absolutely we'll try to do it again uh it's always it's fun to mix up the voices that folks here on this especially if it's a voices as informed as yourself carrie so we appreciate you taking the time to share your thoughts with us and the audience uh anything else chris before we wrap up no i just appreciate gary coming on gary hope to see you in person here sometime again soon and um i've been trying to twist gary's arm and teaching me how to fly fish so i can so i can be ready to retire here soon but um hey i'm ready these days i am ready we just gotta plan on a weekend to meet somewhere in this yeah yeah i'm just gonna have to come down to davis we'll go to the uh american river or uh what's your other one that you like to go up to shasta right um well the sacramento river upper sacramento river above shasta's a river i fish a lot for trout but then i fished a lot up in the mountains and when you said american or i fish the small streams way up in the mountains a lot per trout east fork of the south fork of the middle florida [Laughter] good good good all right well awesome thanks again gary appreciate your time and i will give one last shout out here uh my gonzaga bulldogs have a big game tomorrow so uh go zags let's get it let's get that title this year all right then let's do oh man it's about time all right awesome well this has been full momentum in hec raz podcast thanks again everybody you

Channel: The RAS Solution

Views: 4,420

Keywords:

Id: 4-bQa8Vvoh0

Length: 88min 51sec (5331 seconds)

Published: Tue Mar 29 2022