When David Werth started this lab over five years ago he literally bet the house by maxing out his credit cards and getting an instrumental loan to start the lab. His bet paid off. David is one of those guys that is full of passion when he is describing the lab. This is one of only a few labs in the world that conducts hundreds of physical hydraulic model studies for a variety of water intake and pump stations.
So what we do, is we physically model systems with high turbulence to where if we get that turbulence estimation wrong, it would have a big impact, by doing a physical model we have the real turbulence in the real world. We have real water, we have real gravity and we have real viscosity.
Dr. David Werth
This episode explains why physical models are still important, why a dimensionless number holds a secret to scaling a model, and why water doesn’t always behave the way we think it will.
Transcript
Speaker 1:
Outfall podcast, exploring the hidden edge of our natural water world and our infrastructure.
Robert:
Welcome to the Outfall. This is Robert and today we talk with David Werth, who is founding partner and principal engineer of a hydraulic laboratory located in Anderson, South Carolina. This is only one of few labs in the world that conducts hundreds of physical hydraulic model studies for a variety of water intake and pump stations, including those for water, wastewater, cooling water, and seawater applications. In any given week, they’re working on between 25 to 30 structures, which means they’re constructing, testing, modifying, or demolishing scaled down models every day. When David Werth started this lab over five years ago, he literally bet the house by maxing out his credit cards and getting an instrumental loan to start the lab. His bet paid off.
Robert:
David is one of those guys that’s just full of passion when he’s describing the lab. Hold on to some. Today you’ll hear why physical models are still important. Why a dimensionless number holds a secret in scaling a model, why water doesn’t always behave the way we think it will. And we’ll even get a glimpse into the future of hydraulic modeling. After the interview. Amy and I stop into [inaudible 00:01:34] Dr. [inaudible 00:00:01:41]’s office in Clemson to share a few thoughts on our interview.
Robert:
So, are we still need physical models in this day and age? I’ll give you a hint, and it’s one word, turbulence.
David Werth:
Physical modeling simply fills a niche where the CFD isn’t as strong. And so the best way to kind of describe that is CFD, Computational Fluid Dynamics solves a set of three complex equations and part of those equations has a turbulence term in it. And that’s difficult to quantify. Turbulence is hard to quantify, so to get those solutions to close and reasonable times and with reasonable computing power, we often have to simplify and approximate that turbulence term a little bit. And that’s typically fine. In the Water world, things like clarifiers or if we’re going to put some effluent out into a body of water and we want to see how it disperses, those are relatively stable, calm environments and they don’t have a lot of turbulence. And so if you have to approximate that bit of the solution, it’s no big deal if it’s not perfect, right? In the structure you just see and that you can’t see on the radio here but you guys see looking out the window, those are very turbulent structures.
David Werth:
Those are full of turbulence and so if you approximate the solution in one of these and it’s not quite right, it can have a big impact on the results of the answer. So what we do, is we physically model systems with high turbulence to where if we get that turbulence estimation wrong, it would have a big impact, by doing a physical model we have the real turbulence in the real world. We have real water, we have real gravity and we have real viscosity. So, the turbulence takes care of itself, if you will. If we’re looking for a system where we get that a little bit wrong and it has not much of an influence, maybe CFD is a better solution for that.
David Werth:
We also do a lot of hybrid modeling, where we mesh the two of them together. And if we’re going to model, say the Mississippi river, although there are places where we have physical models of the Mississippi river, but that’s a good example of a very long body of water to where maybe we want to look at one particular area, a lock for example in detail. And we want to determine what’s the influence of that lock on locks upstream and downstream. We might model areas up and downstream of the one lock in particular with CFD and get some boundary conditions for the physical model and then make a much smaller physical model that way that we could look at it in more detail or vice versa. Maybe we take the results of the physical model and feed it to the CFD model and they use that for their boundary conditions.
David Werth:
So those are called hybrid models and they’re used in more and more and more all the time. So they both have places, they’re both very important and some are better than others in certain applications.
Robert:
Do you remember any particular moments when you were in the lab and you saw something and said, Whoa, this is what we talked about in class or this is different than what we talked about in class? Sort of an aha moment of seeing something in person?
David Werth:
I still get surprised. That’s why we do these. To do a physical models is an admission that you don’t know something about it. Right? That’s why you’d still do this. We are regularly surprised. In fact, a good example of that, we did a project, I think it was in Hungary or somewhere in Eastern Europe and you walk out to this big power plant and the water level was surging up and down by over a meter. I just slashing up and down, up and down on each side. Nobody could figure out what it was. So we modeled it and we came up with a way to fix it, but then not more than, I don’t know, maybe a month or two months later, another of the same problems showed up in the lab. We didn’t anticipate it. We said, Whoa, this is kind of interesting. So, that turned on to some research.
David Werth:
And so we will often find a problem out here that… And then we’ll turn around because the model has been built and we’ll keep it a little longer and we’ll do some basic pure research on it and try and understand it. And then we’ll write a paper about that. And so we’ve done that with oscillating flow. We’ve come up with some things on those. We’ve come up with some improvements to self-cleaning wastewater pump stations type of things. So, the aha moments come up with a kind of, why am I seeing this problem over and over again? We shouldn’t be, right? In this day and age but yet we are and so water doesn’t always behave the way we think it will all the time. And so those are fun moments typically.
Robert:
All right. So, tell us maybe the portfolio of projects you’ve got going on right now, maybe it’s some of the locations. Where are some of these projects?
David Werth:
Right. So our five big markets, our power, water, wastewater, process and flood control. And as we walk through the lab, you saw one of each of those markets out there. Right now, today. We have a flood control project in India right now, we have a power project in Thailand. We have a wastewater project in Raleigh, North Carolina. We have a flood control project in Florida, right? So they’re all over. We’ve got them on a lot of different continents and for different markets. But they all have similar problems. And so the market, the way you do the model doesn’t depend as much on the market, but the solution does. So, what we would use for a fix, if you will, in wastewater is very different than what we would use for the drinking side of the water system. So that’s what is different between the markets, but the process is similar for all of them.
Robert:
So, all right. So India, how did they find you?
David Werth:
How did India find us, right? So, this particular case, this particular project in India is from a pump company in Spain who we’ve worked for years. And so that pump company in Spain said, we’re concerned about the intake structure. We’re concerned that pumps are going to have problems when you turn it on and we need someone to look at this and make sure that you’re going to be okay. And we know a group that does that back in the States and so they call us up and we get together and here we are.
Robert:
All right. So I’m also curious about the timing of the studies. It seems like a lot of the work is building the model and then you, you keep talking about when you run the model, like how long does the model run? How long does it take to build typically and how long does it take to actually run the experiment?
David Werth:
Most the course, it depends on the study that you’re doing. The bigger the longer it takes, but on average it’s six to eight weeks for a study. And a lot of that time is waiting for a spot to open. So, the things that you saw on the acrylic shop today are for projects that aren’t even out in the base and yes. So all the acrylic work will be done at a wait for a spot to open up. And then when the carpenters come in, all the acrylics finished and waiting for them to go right afterwards. So there, it’s kind of like framing the house. There’s an empty lot one day, the next day the house is up, right? It’s there. So, it’s fast once all the parts are built and ready to go. So, that’s about a six week process. So, half of that’s construction, half of that’s testing.
David Werth:
We do three phases of testing. Our baseline testing, our modification testing and our documentation testing. And baseline is just to run it as the customer or engineer designed it or exist today. And that’s to identify, are there any problems? And if so, what are they? The modification testing is to fix those problems or improve it. And then the documentation testing is we’ll bring the owners in, the engineers and they’ll look, we’ll make our recommendations. They’ll come back and say, that’s great. Or Hey, can we move this to put a ladder here? Can we tweak this over here? Because this particular water is caustic, so we want to use FRP instead of steel or we can’t de-water for concrete. So we tweak these things to help make it economically… If you can’t implement it, it’s a worthless solution. So we’ll try to come up with something that everybody wins with and that might mean a little bit of a change from what we had. So we have another chance to document that change the last phases.
David Werth:
Typical physical model scale for hydraulic structures in the Civil Engineering world is somewhere between say a one to five and a one to 15. That’s a good range for you right there. When you get out into rivers and big things like that, you might be in a one to 100 type scale, but you’ll distort those. You’ll distort the vertical and horizontal scale. In these types of structures, a one to 10 would be a really common type scale but it’s-
Robert:
Do you essentially make it as big as you can for the facility size or?
David Werth:
You do. One-to-one is your ideal model scale, right? Because then you don’t have any scale effects. There’s always a scale effect when you shrink something down. The goal is to shrink it down enough to where those scale effects don’t influence our results. Right? That’s the goal to where there’re a small portion of the solution.
David Werth:
So, to scale the structure depends on what you’re modeling. If you’re modeling a pump station, then we’re going to scale it based on Reynolds number at the pump bell. If we’re going to scale say a filter membrane, right? Then we’d be looking for surface tension on the filter. That [inaudible 00:10:37] at Weber number for that. If you’re trying to scale a pressure wave, it might be the Mach number. So it just depends on what you’re trying to scale. And something scale better than others. Some things don’t scale well at all in the model. Transients through a pipeline system, we can’t scale the speed of that pressure wave or the elasticity of the material very well. So, it’s a poor, it doesn’t scale well at all. So, some things do not scale well on the lab, something scale with almost no error whatsoever. They’re done properly.
David Werth:
So they’re all function of what you’re looking for. You non dimensionalize that thing, that force. In open channel flow, gravity is what makes things move. So we scale gravity to the velocity of the disturbances. And I’ll get just really quickly into that. If you throw a pebble in a pond and you get a ripple in all directions. That’s called a [inaudible 00:11:22] wave, a pressure wave. That’s a disturbance and water will take the path of least resistance. And so you need to replicate all the disturbance or resistance in a flow field. So, if I have a bunch of piers in a river, like bridge piers. Water hits them and creates these ripples, right? They create a pressure wave and water has to go through that pressure wave or around that pressure wave. So, those scale according to Froude similitude that’s the velocity over the pressure wave disturbance.
David Werth:
And so, by replicating that path of least resistance in an open channel flow, that path will follow the same path in the model as it does the prototype. So, it just depends on what you’re looking for, how you do that scale.
Robert:
So educate our listeners on what is a Froude scaling, exactly?
David Werth:
Okay. So, a Froude number, it’s like a circle. A good way to describe this as like a circle. Everybody knows what pi is. When we figure out the area of a circle is 3.14, right? Now, the units of pi, right? What are the units of pi? They didn’t have any, right? It’s 3.14, so it’s called a dimensionless number. So we can use a dimensionless number. It’s very powerful. So, I could take a circle and if I know what the relationship of variables within a circle. So, pi is a relationship between the area and the diameter or the area and the radius.
David Werth:
So, if I know how much the diameter changes, because I know the relationship between the diameter and the area, I can predict how much the area will change, without measuring it. Okay? So, that’s a pi term. The Froude number is a pi term. It’s a relationship just like 3.14, but instead of being a relationship between the area and diameter, it’s a relationship between the velocity of fluid and the speed of disturbances in that fluid is what that is. And why that’s important is, because to move upstream in a flow, the disturbance has to move faster than the water. If the disturbance moves slower than the water, right? It gets washed downstream. So, all the water coming never knows that disturbance is there. We would call that super critical flow, so the Froude number would be greater than one. The speed of the water is greater than the speed of the disturbance.
David Werth:
In subcritical flow, then that speed of the disturbance is faster than the speed of the water. So, it can go upstream. It can move upstream and tell the water that’s coming. Hey, there’s something down here, right? There’s a volkswagen in the river here. It’s easier to go around me than it is to hit me every time. So, that would be subcritical flow, downstream controlled. So the Froude number is a dimensionless relationship of those two. So, it allows me just like a circle to predict how a bigger structure will behave if I measured it the smaller one. Just like a circle. So, back to my circle analogy. I can take a circle and I can count how many marbles I can fit in the circle. I can predict how many marbles I can fit in a bigger circle if I know how much bigger it got, because I know the pi term for it. That’s how a physical model works. Just like that. It’s really quite simple. Right?
Robert:
On our tour, you’d mentioned something that… You had said, this is a little bit of art and experience.
David Werth:
Sure.
Robert:
Which is kind of funny because as I’m sitting here, you’re mentioning the word art because I’m looking on a board right now with lots of equations on here. Where does the art come into all this?
David Werth:
Oh, there’s a lot of art and physical modeling. Interpretation of your results, right? So, I had a professor that was fantastic. He was a sediment guy. And he used to say there’s far more art in sediment than there is science when you do that. Because it’s just so, so many variables that aren’t well-defined and how they interact with each other, the interaction changes and it’s just not… It’s like turbulence, right? It’s not well defined. So, in these structures, when you see turbulence and you look at these things, you have to remember that what looks like a tiny disturbance in a small scale model can be a much larger one in the prototype. So you have to constantly think in a different scale than you see, right?
David Werth:
And so you have to be able to visualize something different than what you’re seeing in front of you to understand if it’ll be a problem. Some things just simply don’t scale well like air bubbles. So, air bubbles in the model… So if we do a drop shaft structure in Charleston for example, and we’re dropping that water down 30 meters and it’s going to plunge and drive lots of air in there. And if that air gets into the horizontal pipes, it’ll blow back and off core manholes, right? And killing tourists is a bad thing.
David Werth:
So we don’t want manholes flying around. So we can’t have air getting in our drop shaft structure. So we model them and we look at the air bubbles and we say, boy, look at all those air bubbles. They’re going right back out. They’re not going to get in. Everybody’s really happy. But what we forget is the air bubble in the model and the air bubble in the prototype are the same size. And because they don’t scale down. And so that air bubble in the model rises at the same rate as the prototype or the real structure. And the model is only a 10th of the size. So, it gets out very quickly relative. So we have to say to ourselves, how would I interpret that? Right? So, we might say, well, if air gets halfway, that’s we would say it gets all the way in the real structure, right? So we have to apply a little bit of… It’s not an inaccuracy as much as an interpretation. So…
Robert:
What do you think the challenge for you kind of going forward in the next five years?
David Werth:
How do we best meld and integrate CFD and physical modeling together? That’s the future of this right there. We don’t commercially offer CFD out of this facility. Other labs do. There are some great guys that do it. They’re very, very good. It just doesn’t kind of in our type of business model here, we’re not doing that right now. When it makes sense for us, we’ll absolutely do it. We’ve got the data and the expertise to do it. So, we’re looking, always looking for how we can meld those two together cost-effectively. Because it doesn’t make sense to do CFD on something that you repeat with physical modeling. Right? But how can we make them both to where we minimize the amount of both of them and yet we get to the same result and nothing’s repetitive. So that’s what we’re looking for. That’s the next thing. Going entirely with CFD isn’t going to be the way to go for a while, but absolutely hybrid modeling together is.
Robert:
When do you think that only CFD will be? What year?
David Werth:
I don’t know that answer to that question. It’s…
Robert:
Within our lifetime?
David Werth:
Oh, I certainly hope so. One of the limitations is the cost of the modeling and the computing power. The good codes, the Fluents and the Flow-3Ds and out there, those are expensive. Those are expensive seats to have, right? They’re better in the academic side where you get discounted seats. But on the consulting side, companies can pay hundreds and hundreds of thousands of dollars to have access to the software. And so when that becomes a… We’re working on it every day with the Hydraulic Institute. When that becomes, I guess if we want to say more open source, but reliable. Opensource is risky because if it gets changed and you don’t know the change that happened, it’s a risk. So, when it becomes a little bit more accessible, I think we’re going to see bigger leaps and bounds with it and I don’t know when that’ll happen yet.
Robert:
If you just want to reflect, what do you think maybe the one or two keys of your success has been, in the last 20 years?
David Werth:
Well, I love to teach. That’s what got me into the classroom to begin with. And I will, as I wind this career down, go back and teach high school physics.
Robert:
Get out.
David Werth:
Yeah, that’s my plan. I’m going to teach high school physics.
Robert:
Really?
David Werth:
Yep.
Robert:
So, why?
David Werth:
I love to see the light bulb go off. So one of the keys, one of the ways to succeed in this particular business is to educate your clients and [inaudible 00:19:24] owners why this helps them and why this is good for their customers? And so everybody that comes here is subjected painfully to an hour or more of lecture of me [crosstalk 00:19:35] . And so, I think everybody that leaves here, I hope they learn something about hydraulics. So they’re a better engineer. And what I want from them is to not need me for every project, but then they’re going to know when they really do.
David Werth:
And so that’s an important part of this business because if you scare everybody into needing a model for everything, eventually they’re going to realize they don’t. But if you say, listen, these 50% of what you’re doing, there’s no need to model this, if you think about it this way. Then they’re going to know, Oh, we can’t think about it that way. We better bring these back to the lab and take a look at them and that’s how we’ve done well. That’s how almost all of our work is repeat customers.
Amy:
What are the quotes that I took away from this is when he says to do a physical model is to admit that you don’t know something and he takes people’s questions, people’s problems, people’s concerns. How many millions of dollars will I lose if this goes wrong? And he tries to figure out, will it go wrong? How do I fix it? How do I make it better? He distills down issues into workable bite size pieces that you can assess and figure out what went wrong, what went right and how do we fix it?
Dr.:
Yeah, very cool. I think he’s the design engineer’s friend because the design engineer comes up with a plan, comes up with a design, and then they take their precious design to him and he actually builds their design in small scale and then test it and says, okay, your design is great. It should be awesome validation for an engineer, or it needs some tweaks. Then he knows how to tweak it.
Robert:
One of the things I got away with this is that, when we talk about physical modeling versus the CFD modeling, computer modeling, right?
Dr.:
Computational Fluid Dynamics.
Robert:
Exactly. But physical modeling, you think would be really expensive, right? You’re building all this stuff. It’s very hands on. Right? But what did he say he charges? It’s $35,000, $40,000 for a physical model. And when you’re looking at that, okay, spending $40,000 to figure out if this $10 million intake structure is going to work.
Amy:
You think about the information that you’re gaining from that. Almost priceless.
Robert:
It is.
Dr.:
Yeah. Yeah, he talked about that being a small drop in the bucket for the insurance that you’re buying to make sure your design is going to work.
Amy:
Absolutely.
Dr.:
But you bring up a good point about the comparison between computer modeling and physical modeling. Computer modeling has come a long way, but there’s still a lot that it cannot do and I think Amy, you had a thought about that too, [crosstalk 00:00:22:51].
Amy:
Yeah, when he was talking about the physical model versus the computer model. With the physical model, you have all of your dimensions right there in front of you. Something’s wrong. It’s going to pop out. It’s going to scream at you, but with the computer model, you have to look for a problem. You have to go get each one of the dimensions and know what you’re looking for and know what problems you might be able to anticipate and fix. But the physical model, you are going to make sure that you see every dimension you see every problem.
Dr.:
Yeah, yeah. That’s cool. I think there’s a lot of things in life that are going to be that way, that even though we’re doing so much digitally or virtually, that there’s still going to be a place for the real.
Robert:
Analog, vinyl records,
Dr.:
Right. There you go. Vinyl records. People still love that sound of the vinyl that you just can’t capture.
Robert:
[crosstalk 00:23:52] say it sounds better.
Dr.:
Another place is just conversations, like the three of us are here face to face, having this conversation.
Robert:
It’s awkward.
Dr.:
It’s a little awkward, but there’s just something to being face to face that you can never capture virtually. And I think this is why we still have conferences, even though we could call each other on the phone or use Skype or whatever, we all still like to get together physically in a physical space to converse. And I don’t see that ever going away. I hope it never goes away.
