Ep 45 - Building an LSS Computer Model

[00:00:05] Greetings all, Joe Arlotto here with Dr. Jeff Kefaber. Welcome to Inflow, the podcast where we discuss all things related to aquatic animal life support. The show is sponsored by Dry Naqua and Asahi Valves. You can learn more about our sponsors by clicking on the sponsors page at inflowlsspodcast.com.

[00:00:27] Innovation, as in life and technology in general, so does LSS innovate. Our sponsors today, Asahi Valves, Chris Kane, they're always thinking of the next valve, the next pipe material. Mike DeMaine at Dry Naqua, developing new filter media.

[00:00:52] So, you know, someone's always trying something new, a different filter, improving pump operation, designing a better process. What does innovation mean to you, Dr. Jeff? Innovation, I think about almost exactly 10 years ago this week when a group of consultants came together to talk about an idea and that idea was SeaWorld Abu Dhabi.

[00:01:15] And the challenge was placed before certainly me that we need the next generation LSS design for this park, for this new next generation park.

[00:01:29] And so that, you know, the challenge was thrown out there and that required high level thinking about the processes design that is not your typical, you know, cookie cutter type LSS that we've been creating over the decades.

[00:01:48] And so there was that, that innovation. And then later, whether it was San Antonio or even Abu Dhabi, where we were introduced wetlands and salt marsh filtration into LSS, that was certainly innovative. Yeah, it's a very good point. Very, very good example that you bring up with SeaWorld Abu Dhabi.

[00:02:09] You know, I think about when I first started in life support, you know, and all my systems had airstones, large concrete towers with a manifold on the bottom of 20 or 30 airstones where we bubbled in the ozone.

[00:02:26] And those bubbles were small. I didn't think it needed to be any better. I mean, I'd have to dive the tower to remove the stones, to replace them or clean them.

[00:02:39] But when I got word that there was a better way, you know, through the industry, through my contacts, they said, hey, you can use these injectors, specifically MAZI injectors, which were originally developed to inject fertilizer into the water for crops. So we retrofitted our systems with these injectors and ozone contact was more efficient. The bubbles were smaller and the system itself required less maintenance.

[00:03:09] That's exactly right, Joe. The injectors produce a high surface area micro bubble, which dissolves into the water much more efficiently. And to take this one step further, injectors are now employed not only in traditional ozone contact chambers, but now pressure contactors, as well as fractionators or protein skimmers, where we inject ozone directly and make them the contactor.

[00:03:37] What used to be a crazy innovation borrowed from another industry, you know, is now the LSS standard, right? That's exactly right. I mean, what happens when you have new ideas and innovations come into play at whenever they're a part of the process, they become a gold standard and they become repeated over and over again.

[00:04:04] And they become a artifact of new LSS design. Yeah. So the also conference, which we've talked about a lot on the show and that happened in Kansas City last month. So the also conference is a great place to learn about these latest innovations. Dr. Jeff and I met up with Paul Cooley from Ardura. Remember, we had Paul on the show very early on last year.

[00:04:30] He's an LSS engineer who's been doing this for a very long time, really one of the first who did it, you know, exclusively. So we sat down with Paul at also to talk about, you know, his latest innovation. And you'll have to excuse my voice during these interviews because I was losing it from talking so much at also. But here's Paul. Paul Cooley. How are you doing? Doing good. So we're here at the Alta Show and catching up with Paul.

[00:04:59] And here we get to see a lot of people that we work in the industry with and we get to catch up on the latest technology and see what's working on what. So what do you think about this day, Paul? Well, it's been interesting. And this show is a great opportunity to kind of bounce ideas off people. And since the last episode of Inflow where I talked to you guys, I've been busy.

[00:05:24] So we have been working on doing what we talked about at that session, which was really to try and do a more complete water quality model for life sports. So it's been, I don't know, six months or maybe a year. And we've been actively developing a computer model that can basically predict all water.

[00:05:52] Now, it's right now it's focused on water quality, but I've got some interest in actually expanding it beyond that. But we have that model now functioning and we have tested it at Long Beach. So we use the Long Beach water system as sort of a proof of concept. And so the Long Beach water system is extremely complex because of the way it's actually grown over the years.

[00:06:21] And so it's a perfect way to determine whether the model is functioning or not. Now, you, just to sort of go back a little bit, because Paul was one of our very first guests when the podcast started about a year ago. And Paul is an LSS engineer. So you actually did the design of the original Long Beach Aquarium way back when, correct? That's right.

[00:06:48] And of the original sea honor system for the Long Beach Aquarium. But since then, what's happened is it's actually morphed because they've added fish. So they've got a lot of fish in there. And so it's a combined system. And Paul, with that, on that note, you have added over the years to that system, right?

[00:07:14] There are other loops, other treatment trains, other processes that have been added. And as a test case, that otter habitat and the life support system has a plethora of processes already integrated into it. Is that right? That's right. And so the reason that we actually were working on that system was not because of this model.

[00:07:42] It was to track down DO drops. So there are some problems with dissolved oxygen on that system, predominantly because you have a heavily loaded sea otter system combined with fish. So the DO drops. And they've noticed that this happens when they have power failures. So they were doing simulated power failures and the DO was dropping too low too quickly. And so they wanted to be able to troubleshoot the DO drops. So that was the initial reason for looking at it.

[00:08:12] But we sort of used the model then to track down those issues on dissolved oxygen. And then we expanded it to include the nitrogen cycle, the phosphorus cycle, the carbon cycle, solid balances, and even temperature. So the model has the ability to model all those.

[00:08:32] So a long list of water quality parameters can be modeled and predicted right down from DO to ammonia, nitrite, nitrate, phosphate, as you point out with the nitrogen, phosphorus cycles.

[00:08:51] So is there a subset of parameters, like a small list that you're going after first and then expand the list of parameters? Are you going after everything now? We're going after everything. But the actual use of these models requires validation. You can't just throw a computer program out there and then just pretend like it's going to work.

[00:09:20] So it needs to be validated. We're getting some validation on Long Beach because Chris Carr and the Long Beach staff have been really excited about this. And so we've been able to validate some of the systems, but we're not completely there yet. So there's a long ways to go. But I think we're probably about maybe 70%, 80% to be able to model all the water quality you're talking about.

[00:09:47] But I'm talking about pH, alkalinity, carbonate, bicarbonate, CO2. So the whole carbon cycle, obviously ammonia, nitrite, nitrate. And then the tricky one is turbidity because that's a solid balance. Now, this model is all solid. It's a balance, a mass balance.

[00:10:11] It's basically saying incoming plus accumulation or depletion equals output. It's a very simple mass balance. But the algorithms that are driving it are not simple. They're very complex. And so the idea is to not make this just a, we would like to remove, you know, 80% of the ammonia. And we assume it's 80% of the ammonia.

[00:10:40] But when we change the temperature of the water, we change the nature of the biofilter that it then has to be recalibrated. The idea is to actually measure the nitrification of any type of biofilter to allow you to change the media and the temperature. And boom, it changes the ammonia removal. So that's the goal. It's very complex. And this is a sea otter. And they've added fish to this habitat, correct? Right.

[00:11:09] So let me just follow up on that a moment. When you add fish and mammals in the same water, obviously food load increases. And as facilities add animals, add otters, individual animals to their systems, and that results in a higher food load.

[00:11:35] I mean, don't you really start with the food loading to calculate nitrogen input? Right. Yeah. It actually starts with the food itself. What is in the food? So you feed an otter different than you feed a fish. So there's organic nitrogen in the food. There's lots of inorganic nitrogen. Protein. There's proteins in the food. The proteins break down. As they pass through the otters, they break down. Some of them are ammonia.

[00:12:04] Some of them are organic, Keldon nitrogen. There's a lot of different pieces of it. And so it's very complex to even determine the initial load in the system on the nitrogen side. The solid side is a little bit easier because a certain otter is going to produce a certain amount of solids from a certain amount of food. So you can predict the solid side a little easier on that. But everything has to be predicted. Phosphorus has to be predicted. Everything.

[00:12:32] How might you use this model? Like what would be the advantage of this model is going to be in and predict certain things. How might a life support operator use this model? Okay. So the model, my goal of the model is to have a number of different uses. From an operator standpoint, it can be used to actually create a tank.

[00:13:00] Let's say a $300,000 shark tank. And you can tack the system on it. So you can put a pump and pumps on it. You can put sand filters on it. You can put heat exchangers on it. You can put a fractionator on it. You can put an ozone contactor on it. And then you can push the simulation button. And it basically cranks the simulation flow out of it. And it tells the operator, with this system, I can control this amount of makeup water, this amount of water changes.

[00:13:29] I can now control the ammonia at 0.02 milligrams per liter if I run it this way. And then what would happen is you could take a certain number of fish. And then you can say, what would happen if I double the load? And you could throw a double load on the fish. And then you can calculate what the ammonia levels are and the nitrate levels are. What the equilibrium nitrate levels are. So it's very interactive.

[00:13:54] And you can learn that, you know, if you double the load, you're going to see an increase in nitrate. So you've got to make sure you accommodate that. And so the same thing with solids. You're going to see an increase in turbidity. When you clean the tank, you're going to see an increase in turbidity. The filters are eventually going to connect with it. So it's educational. Then they can go in there and it can say, well, I want to change out a perlite filter, you know, a defender filter. And I want to change a defender filter in there. What does that do to the water quality?

[00:14:23] What does that do to the saltwater discharge? And then you can throw in a drum filter. So you can throw in anything. The goal is to have everything quantified. So you can throw in anything you want and you can calculate the nitrogen cycle, ammonia nitrate nitrate. You can calculate equilibrium phosphate levels. You can calculate the pH, the alkalinity. You can increase the efficiency of the D-gas tower. And you can see the CO2 drive it off. And then the pH drops. So it's, I mean, the pH increases.

[00:14:51] So the, the, the ideas are brought by everything. Now we're not completely there yet. So it's an educational tool for the, the operators. That's one use. The other use of troubleshooting. DO troubleshooting is an example of it. So we, we took the otter system and we threw food into the otters. We threw food into the fish. The, the otters produce certain waste.

[00:15:17] There's a certain amount of biochemical oxygen demand associated with that waste. It consumes oxygen. There's air, there's water, oxygen transfer across the surface of the pool. Calculates that. It pushes the, the water from the pool into the filter. The filter accumulates solids. You can backwash and you can, you can see the DO drop in the filter. As it goes through the backwash cycle, you can backwash more frequently. You can move the solids quicker. Reduces the amount of oxygen demand associated with the sand filter.

[00:15:46] Then you push it into the de-aeration tower. So we chased the oxygen around in the system. We chased it around and we calculated how much oxygen was in the overall system. And of course, the most important thing is where it's in the tank. So we wanted to know how much percent saturation was in the tank. And so we ran it and then we went and sampled in those sample points. And we, we basically validated the model. And then we said, okay, now I want to increase the DO. Well, how do I do that?

[00:16:16] There's different ways of increasing DO. So this model can be used to simulate the, the, the changes in the system and see how it drives the DO up. So let's talk about dissolved oxygen a little bit more. Now, everywhere in the system and in particular inside sand filters, there are, there are nitrifying bacteria that consume oxygen to oxidize ammonia to nitrate ultimately.

[00:16:44] So a dirty, if a sand filter gets dirty, it has a higher BOD, oxygen demand, the nitrifying bacteria also consuming oxygen. So what, but that could happen elsewhere in the system too, right? It could happen on de-aeration tower packing media.

[00:17:11] It could happen on the surfaces in the exhibit itself. Substrate, whatever furniture they put in there. So comment on how you model DO system-wide. Okay. So each process has a pop-up on it. It pops up the tank, for example.

[00:17:36] It asks you to identify the surface of the tank. So it, it knows the water surface square footage. Now this is a relatively small tank. It's a 55,000 gallon tank, but it doesn't really make any difference how big it is. So it can model the surface of the water. And so it can calculate the surface transfer from the air into the, into the water. Okay.

[00:18:01] So then you can then decide that you want the tank to be biologically active. It's a selection. You select it and you click on it and you say, I want it to be, I want it to calculate the nitrification of the tank. So it has the surface area of the tank because it knows the overall surface area, surface and surface, you know, walls.

[00:18:29] And then it calculates nitrate removal for the tank. It turns the tank into a biofilter. You can change it. You can choose to do it or you can choose not to do it. Then you push it into the sand filter. You can choose the sand filter. You can put AFM in the sand filter and you can click on it. You can say no biological activity in the sand filter and boom, it turns it off or you can turn it back on again. So it basically allows you to trigger those along, along the alignment.

[00:18:59] So you can do it in a biofilter. You can put an external wire filter in all of it. So on the platform, on the software, you've got all a menu of choices, right? And you can click on those, put a checkmark there because I want to activate that parameter or turn it off. Right. Right. And it could be that there are certain parameters that aren't there. And so we need to then add them.

[00:19:28] You know, it may be that. So we're not saying that everything is there. We're just saying that the program. Okay. So that, that analysis, Mondial, is just an example of the use of the system. So one uses for the operator training, one uses for troubleshooting and coming up with solutions on a system that's having problems.

[00:19:54] What about the LSS designer or engineer like yourself, Paul? Is it a design tool as well? Right. That's the third use. Well, actually there's another use too, but the third use is the design tool. So we can, you know, we're doing a Columbus Aquarium. We can put in the Columbus Aquarium system. We can play with the different options and try and optimize the system before we build it. Yeah, that was my point.

[00:20:21] If it's a new build and you're designing a system or a whole park or a sea world, for example, in Abu Dhabi, you could use it as a tool to inform that design choice. Right. You know, for example, Abu Dhabi is a good example. So Abu Dhabi is unique in that they have a lot of seawater access. It's it's I mean, you know, the issue, the salinity issues there.

[00:20:50] So you can play with the salinity issue. But if you were to take Abu Dhabi and move it to an inland facility, then you would lose that seawater make. So you can take the technology of Abu Dhabi and move it to a new system and then incorporate in the artificial seawater. On that point, Paul, what about sustainability choices that we want to make as we operate our systems long term?

[00:21:18] Like we want to turn pumps off or turn filters off or maybe we don't chill as much at night or during the day or whatever. And we were trying to save energy and also conserve water. So is this also a tool where I can go in and say I've got 16 sand filters and I'm turning eight off? Yeah. Yeah.

[00:21:46] The the pumps and even the pipes are active nodes. So you can click on a pump and you can you can determine the characteristic of that pump. You can click on a pipe and you can make the pipe smaller or larger. And then once you're running, you can turn pumps off or you can ramp them down and reduce the power costs. And then it calculates the power costs. It's calculating the total energy being put into the system. And so it calculates the chilled water requirements.

[00:22:16] It calculates the hot water requirements. And so you can save energy by turning stuff off and you can optimize that. And I think that's a very good point because we've talked before about how a lot of these systems are overdesigned. They're not. That's not an evil thing. It's just that we don't anticipate. We anticipate a certain load and then the load doesn't occur for whatever reason. And so you should be able to do some energy savings on this. But the opposite can also be true.

[00:22:44] And depending on the facility, we add more animals than was designed for. And as we go to multiple species exhibits and we add fish and invertebrates, that puts more demand on a sea otter or walrus type system that has already got a high load. Yeah, the Long Beach system is kind of an interesting example, right?

[00:23:11] I mean, that thing started out as a very basic sea otter system. And then all of a sudden they wanted to put fish in. So, you know, God bless the Long Beach Aquarium. But, you know, they had to accommodate that change in the system. And they did it the way they thought was most efficient. But if we would have had this program, we could have anticipated the water quality change and what the DO drops were going to be. So maybe we could have made the renovation for fish installs more efficient.

[00:23:41] That would be one use of it. And as I understand it, these are very large yellowtail fish. And so they do have an extra demand on that system, not just DO, but ammonia excretion. Right, right. So one is operator training. I'm very interested in that. I think this would be a very useful tool.

[00:24:04] Second would be troubleshooting, diagnosing what problems are occurring and how we can fix them. Third one is design element. So that would be another piece for it. Now, I've got, you know, crazy visions for this thing. Some of them may just be crazy visions, but some of them may occur. Because I want to be able to monitor other things, too. I want to be able to monitor, for example, trace elements.

[00:24:33] You know, instead of doing a water change because you just want to do a water change every, you know, 30 percent every two months or something. You could be trace, you could be tracking trace elements and determine when you need to do the water change to replace the trace element. Or you just replace the trace element. You don't do a whole water change. So the analysis of the water changes could be very functional. Then I want to be able to track the biology of the system using this system.

[00:25:02] Right now, the system is tracking chemical components, you know, carbon and nitrogen and phosphorus. It does not track bacteria, you know, E. coli or enterococcus. But I think we can input a biological component and track E. coli, track certain components. So what's the next step? Like, obviously, to make this model, to make this computer model really functional, you need a lot of data, correct? Yeah.

[00:25:30] It depends on the piece that you're using. For example, temperature control. I don't need a lot of data for temperature control. I can use the program to calculate the chilled water requirements. Now, I'm going to validate it against some other methods, but I don't need a lot of data to do that. So that's pretty simple. Nitrogen control, the nitrogen cycle is pretty well understood.

[00:26:00] And so I think I'm going to look at it more closely, but I think the nitrogen cycle is pretty well done. And there could be some variations once you hit unusual biofiltration. You know, it may not be in there, so I know it'll trigger that. Foceros is an interesting little dilemma because we're finding that the phosphate levels are lower than what we'd anticipate from the model. So what it triggers is sort of an inquiry into where's it all going.

[00:26:31] So the model can track that, and then we can tie down where it's really going. And then, of course, the ultimate question is how much lanthanum would you need to get the phosphate down? Because the phosphate level on this water system is up to six, seven milligrams per liter. We don't really care because it's an indoor tank. We don't care about algae growth or any of that stuff. But on other tanks, we don't care about it. So that requires some more data.

[00:26:57] The carbon cycle is an interesting one because a lot of the systems we're dealing with, we're not all that obsessed by the carbon cycle. We're not obsessed by CO2. Now, some people are obsessed with CO2, and they want to control CO2. I have actually been criticized in the past for not being obsessed on CO2 by certain people who will go unnamed. Okay.

[00:27:22] But the idea of where the whole carbon cycle is occurring, that's of interest to me. But is it really affecting the design? I think it doesn't affect the design that much, but it's interesting. And it may be a troubleshooting CO2. So we may have to add more data on that. We may have to collect more data on that. So there's pieces of what we need to collect, but it's not too excessive.

[00:27:50] The turbidity one is the most difficult one. And that's the one we're going to really focus on is turbidity. Paul, one thing I think about a lot is carbon capture. And as I start to design more and more wetland filtration systems to replace sand filters or traditional methods, we're seeing that the wetlands put down a lot of soil.

[00:28:18] They actually make this real rich black carbon-based organic soil. And wetlands, they start out with no soil. Basically, it's a hydroponic type system.

[00:28:36] So as the wetland over time, over the years, puts down and captures carbon and removes that carbon from the system, how can you model that in more detail? Do you think you can do wetlands at some point with the problem? We want to do wetlands, obviously. We need to be able to do it.

[00:29:03] It's very simple to do it initially. It's very complicated to do it real analytically where we get into the biology and the chemistry of the wetlands itself. So I haven't looked at it that close, but we definitely need to include. I would start out by just saying you're removing a certain percentage of the carbon. You're removing a certain percentage of the ammonia. The ammonia is coming in, or the nitrate in this case. Nitrate is coming in at, you know, 50 milligrams per liter or 20, whatever it is.

[00:29:32] And then it's leaving at zero. So you're seeing a certain percent removal of nitrate. That's how I would do it initially. But eventually you've got to get into the biology and the chemistry of the wetlands itself. Now, you may know more about that than I do. And there may be that data may actually be accessible. So, yeah, I think there's a lot of literature data already out there in the peer-reviewed literature.

[00:29:55] I want to pivot real quick to animal health because at the end of the day, it's what's going on with the animal, their immune system, the blood chemistry. Is there a way veterinarians, and I want to kind of make sure that our animal care professionals are linked in to this technology, this model.

[00:30:23] Are there blood or physiological parameters or measurements that could one day make it into the model? You mentioned E. coli, enterococcus, some of the microbiota that's going on inside the water. But what about the animal itself? Yeah, you know, I think in order to get to the end product, you need to include animal health somehow. But it's very complicated.

[00:30:53] I mean, the issues between specific animals, you know, it's very complicated. And, you know, it's going to be hard to imagine that this would really be possible to be used in that regard. But it's possible. I was thinking that you could do, for example, simple things like putting a camera on a tank and use face recognition software to basically identify specific fish and then be able to monitor their behavior.

[00:31:23] So they're flashing or they're, you know, they're acting something out of sync. But of course, you're not doing blood work on the fish. And so, I mean, I was thinking about that. I don't know. I know they've used face recognition software to analyze microorganisms like protozoans and bacteria. So instead of testing for the bacteria or the protozoan, you basically use face recognition software

[00:31:52] to identify it physically. So I know there's ways of doing it. So there's no doubt that you could use face recognition software on fish and animals, dolphins and stuff. I'm pretty sure you can do that. I've never seen it done, but I'm pretty sure you can do that. So maybe that would bring fish animal health into the equation. But as far as doing blood work and stuff, of course, you've got to get the number into the program. You've got to somehow get a number into it.

[00:32:20] So that's more complex. I mean, in systems that are ultra high nitrate, for example, even a mammal might present with a higher nitrate level in the bloodstream. And I believe it's known that nitrate can interfere with iodine uptake at the thyroid.

[00:32:42] And so there may be markers, chemical markers inside the blood that could work their way into the model at some point. Are you going to continue working at Long Beach? Are there still more questions to ask and more problems to solve there? Oh, yeah. Yeah, we're going to continue working on it. We're going to continue looking.

[00:33:11] We actually have another series of testing protocols that are in place. We haven't triggered them yet, but we're getting set to trigger them to identify some of the holes that weren't filled by the first testing. So we're going to do that. And then we're talking to Long Beach about possibly modeling the whole aquarium.

[00:33:31] So once you establish a template of a certain type system, it's very simple to move it into a different tank from tank to tank to tank to tank to tank because there's not that many different variables in there.

[00:33:45] And so the idea would be to model the whole system and then see how that can be used to do things like energy savings or reduction on makeup water, water changes, stuff like that. So, yeah, no, we're going to spend more time doing that. And here at the also show, have you been sharing this with the folks here? Yeah, I gave a presentation on this yesterday.

[00:34:14] I think it was well received. I think there are some skeptics, which is justified. It's justified being a skeptic. You know, these computer programs can, you know, get out of hand. But I think we're moving into sort of an AI world, right? I mean, we've got to get into that kind of world at some point. And a failure in, for example, the carbon dioxide prediction of this model doesn't invalidate the nitrate prediction.

[00:34:43] So to me, it doesn't matter exactly. Each system is separated. So we can use it where it makes sense and where it doesn't make sense, we don't use it. But the idea is to get into the, I call it the AI world. This can become the algorithm by which you can analyze a lot of stuff if you give it access, you know, to an AI system or not. So, I mean, the potential is crazy potential.

[00:35:09] Yeah, and it doesn't have to be, to be useful, it doesn't have to be 100% accurate. Right. You know? Exactly. Yeah. And there may be cases when you need it to be accurate in a particular area. And so you want to validate the testing in private. And there may be other cases where you can run it and then you just throw a safety factor on and that covers the uncertainty.

[00:35:35] I mean, there's different ways of making it work as long as you recognize the accuracy and inaccuracy of it. You got to make sure you understand it, you know? So. To me, it's kind of the next thing. And why not? You know, why not learn more about our systems? Why not collect more data and help us predict what's going to happen more?

[00:36:00] Yeah, I think about AI and how it's going to affect LSS in the future. And, you know, I do not see AI replacing our jobs, first of all. But this model, Paul, will help operators make smart decisions.

[00:36:24] And if AI can help us be better operators, better designers, save energy, save water, be sustainable, then we're all for it. Right? Right. Exactly. And I really want to give you guys credit for a lot of this because these ideas kind of came up in the initial inflow discussions and it triggered discussions in the industry. You know, I don't do all this myself.

[00:36:53] I'm not a programmer. So it triggers some information. And I really think it demonstrates the value of what you're doing on inflow. It really does. And so I appreciate, you know, what you guys have put together here. Well, thanks. We appreciate the support. Paul's a sponsor. Ardura is a sponsor.

[00:37:16] And, you know, I look forward to this is a good sort of setup for us to talk again. You know, in three months, four months, six months, collect some more data, maybe even work with some other facilities collecting data. And we'll touch base again, for sure. Yeah, right. Exactly. Because I think this is going to be a process.

[00:37:43] And so the more we can identify issues and then apply this and reform it, refine it, the better reality. And we talked yesterday, Joe, and you said it. The more iterations of this, all that data from multiple facilities, each iteration, I would think, would get a model improving all along, right?

[00:38:13] That's much closer. Yeah, and it's not just the programming either. It's the interface. So in order to make this thing useful for an operator as a training tool, the interface input and output interface needs to be friendly. It can't be a bunch of goofy engineers. You know, it needs to be more friendly. So there's a lot of improvement that needs to be done. Great. Well, thanks, Paul. We appreciate seeing you as well.

[00:38:40] I'm looking forward to catching up soon and hearing more about this. Yeah. Thanks, Joe. Thanks, Jeff. Thanks, Paul. Let's take a quick break here to talk about one of our sponsors, Longhorn Organics. They're an LSS installation contractor. We've had Longhorn on the show, Holly and Bo Dempsey.

[00:39:01] Typically, for a new LSS project, you know, it's my experience that the general contractor might use a mechanical contractor to install the LSS.

[00:39:15] I mean, I've worked with some very good mechanical contractors, but many times they don't understand some of the nuances and intricacies of LSS, like high points and pipes or gravity discharge from a foam fractionator or protein skimmer. Or they don't understand sort of the process of ozone degas and destruct.

[00:39:42] But Longhorn Organics, Holly and Bo, they've actually worked at aquariums. They've turned valves and they've backwashed filters so they understand what it takes to design an LSS. And so now with their company, Longhorn Organics, they focus really on LSS installation. You've worked with them, right, Dr. Jeff? That's right, Joe.

[00:40:08] I worked with them in San Antonio at SeaWorld on our LSS design innovation, which included drum filters, perlite filters, ozone fractionator, and a salt marsh wetland. And Bo and Holly were instrumental in that installation. Longhorn Organics led that installation, working with the GC and the owner, me, for example.

[00:40:38] They made some great innovative steps in the design, in the field, and we were able to pull together a system that is unique in the industry that does not include sand filters. Yeah, see, they have that kind of experience. Another thing is bringing them in early in the project so they can provide input.

[00:41:01] Sometimes my experience is that in the design process, the LSS operator doesn't necessarily have a lot of time to be involved in those design meetings. So, you know, someone like Longhorn, if they're involved early, they have a lot of experience. They have a front row seat, really, working with many different zoos and aquariums.

[00:41:27] And they can offer perspective because they've had different approaches to LSS. They can offer a very unique perspective on how to build it and how to plan for your LSS to be efficient through its very long life. So they can help you solve potential LSS problems before construction. We really appreciate Longhorn's support and sponsorship of Inflow.

[00:41:56] And we also really appreciate their commitment to LSS. So Paul's always thinking about the next thing. I mean, many of us are when it comes to LSS. Designers, engineers, equipment manufacturers are trying to come up with better ways to increase efficiency, have better water quality, save water, save energy, better animal health. Right, right, Dr. Jeff? That's exactly right.

[00:42:21] And I think about the sustainability piece is also innovative in the sense that you just pointed out. Paul, for example, has been decades of driving innovation in design. And so he represents kind of one of those founding fathers that we talked about early on. Yeah. And, of course, operators are involved in this innovation, too.

[00:42:50] We need them to innovate these LSS. So we got a chance to catch up with Chris Carr from Long Beach Aquarium of the Pacific. And he's working with Paul on this model. He's providing the platform, which is key. Chris had a lot to say about the model. I know you've worked with Paul Cooley in the past. He was sort of the lead designer for the original aquarium.

[00:43:17] And I know you've made a lot of changes over the years. And sometimes if those changes required design work, you'd probably work with Paul. So we've met up with him at all this year and talked to him a lot. And he told us about some of the things he's doing at Long Beach now. Can you sort of walk through that a little bit and maybe tell us why we're troubleshooting that honor exhibit and what's been happening with it? Sure. Yeah.

[00:43:47] We really love working with Paul. I've worked with Paul for a long time. He's really easy to work with. We kind of speak the same language. So it makes it great to kind of go back and forth and bounce ideas off each other. We were having issues with our sea otter exhibit where the dissolved oxygen levels were running low. And we were really concerned with power outages and how fast the DO would drop in our exhibit. Now, the otters exhibit has large yellowtail fish in the exhibit.

[00:44:16] So those fish would suck up all the oxygen pretty quick within about a half an hour. And we were thinking, well, we need to put one of these three filtration loops from otters on emergency power so we can continue to keep this going. Well, we needed to make a choice where it was just going to be maybe one pump because we didn't have room to put all the pumps on emergency power. So we wanted to determine which would be the best loop to keep going. And in the moment, Paul said, hey, this is a great idea.

[00:44:45] So we said yes. We got together with Paul. And he said, hey, I need you to sample all these different areas in the system. I'm like, okay. He pointed them all out. We walked around and we got samples from the suction side of the pump, the discharge side of the pump. And, you know, before the tower, after the tower, before the degas tower, after the degas tower, all these different places. And it was just a whole bunch of data. And we had also trended data from our Siemens building technology that I, you know, got reports out to them.

[00:45:13] And they could take all that data and put that in their model and determine what happens with our system when something would happen. And it would say, you know, if this is going to be this flow, then this might happen. So it kind of predicts what's going to happen. And it tells you, ultimately tells you what you need to do to fix the problem. It's an amazing tool.

[00:45:34] So, Chris, you literally, with Paul, went process by process through the life support systems, all the different loops. And you did before and after process measurements for dissolved oxygen. So what did you see? Did it help you inform the backwash frequencies of sand filters better? Yes, absolutely it did. And I wasn't even ready for that.

[00:46:01] I mean, we were just looking at DO and it gave us so many great ideas. We were talking about, yes, possibly all the suspended solids inside the filter might have been affecting our DO levels. And we determined, hey, do we need to do more backwashes more frequently? And how much is that going to help us? And it actually showed us. And so it made me realize, hey, I could do one more backwash or maybe it's not going to work.

[00:46:26] And this will work for us because if I do too many backwashes, it's using a lot of operational time and it's using a lot of water and we're not saving that much on oxygen. But that model was great because it told us exactly what we needed to do. So I asked Chris some specifics about what he was measuring. Was he measuring DO levels at various parts of the system so he could determine where it was increasing and decreasing and then putting those numbers into the model?

[00:46:56] And did it help him predict where he had to increase the DO so that his system would give the kind of numbers that he wanted? Yes, actually, specifically in the system. And it's amazing when you see the roller coaster of DO and it comes, you know, the levels on ins and outs of all the different equipment. And it points all that out and it really lets you know, all right, well, this is exactly happening in this part. So what happened? Like what did it tell you to do?

[00:47:23] Ultimately, it told me that we needed to, well, we needed to add oxygen straight to one of the loops. I think it's called the blue loop. And we're going to add an oxygen generator and I think a space cone on there. And we're going to put the pump associated with that on emergency power. And then I think we're going to be just fine when it comes time. Next time around, we have an emergency where the power goes out. When will that happen? When we least expect it. Usually on a Friday afternoon.

[00:47:50] I don't know what the power outages are, but we're getting all our equipment together now and we're going to do the install. I'm very excited to share the results when we get everything installed. I'm staying positive that this is going to work really well. And so, Chris, I think you have a fractionator on one of the loops on that sea otter exhibit that we're discussing for the model.

[00:48:12] Do you feel that injecting ozonized air into a fractionator would then oxidize dissolved organic matter and create the foaming? That you would have a higher dissolved oxygen residual coming out of that process? Right. Yeah, that's exactly right. Yeah, it's amazing to see the differences coming in and out, especially things like protein skimmers. And some of it was surprising. I thought the DO would be a lot higher coming out of our aeration tower. And it was surprisingly not.

[00:48:42] And you just look at it and you think, well, that's what it's there to do. So it's got to be working. But to get the actual data and see it in real time and see what happens in that snapshot is just a game changer. So I asked Chris, what's next? Paul's talked about the model. And DO is one of the many parameters that this model can possibly predict. I asked Chris, do you see yourself using it more for other things? I think it would be a great tool.

[00:49:10] And I'm such a nerd about it. It's going to be, if I get my hands on it, it's going to be more of a toy. Because I want to say, what happens if we do this? And you can just kind of plug and play. And I wanted to predict what my coliform levels are going to be in some of our birds and mammals exhibits. I certainly hope that happens someday. Because I think that'd be very, very helpful. And of course, coliform levels, total coliform counts, is a regulatory thing with USDA.

[00:49:38] And compliance is important. So if the model can help you predict potential failures and help you with your corrective actions, right? Right. Then when the USDA inspector shows up, looks through your records, and he sees, he or she sees what you did in response to a problem, that's valuable, right? It really is. It really is. Yeah.

[00:50:05] So if we're over 1,000 most probable number on coliforms, we need to take action. And that's something either operational, changes in ozone, usually a water change, which could be costly. To be able to predict what's going to happen, say, hey, I need to take my ozone generator down tomorrow. What is that going to do without dosing? What's that going to do to the system? Are we going to fail on Tuesday or whatever it might be? And if I can figure that out, we can plan for that. Plan for that.

[00:50:33] And animals certainly benefit from that. Yes. Oh, absolutely. Yeah. Yeah. So I'm really looking forward to using it as a tool. I'm excited about it. It was really good to talk to Chris and Paul. It's fascinating what they're doing, but the model is only really fully works if you gather a lot of data.

[00:50:57] I mean, Chris is helping do that at Long Beach, but the model gets better with lots of data from lots of facilities. So please let us know here at Inflow if you would be willing to help gather data to help hone in this model. Dr. Jeff, last word. Yeah, Joe.

[00:51:24] And for follow up on that very point, the massive amount of data that we need to further develop the model, you know, first of all, you have to define what parameters you want to actually measure, what data you may already have in your files from years and decades of doing this work. Dissolved oxygen is just one of many parameters. I was thinking about the nitrogen cycle, the carbon cycle, phosphorus, all of the other

[00:51:53] water chemistry inputs and outputs from LSS processes that will help build out this model in the coming months and years. Absolutely true. Well said. Thank you very much to Chris Carr and Paul Cooley. And thanks to Asahi Valves and Dryden Aqua for sponsoring this show. And thank you all for listening to Inflow. Be sure to subscribe and leave us a review.

[00:52:19] And don't forget, you can always stay connected with us at inflowlsspodcast.com.