Numerical Groundwater Computer Models and MNA

Loading...

From the course by Rice University

Natural Attenuation of Groundwater Contaminants: New Paradigms, Technologies, and Applications

103 ratings

Rice University

Natural Attenuation of Groundwater Contaminants: New Paradigms, Technologies, and Applications

103 ratings

Cleaning up the large number of groundwater contamination sites is a significant and complex environmental challenge. The environmental industry is continuously looking for remediation methods that are both effective and cost-efficient. Over the past 10 years there have been amazing, important developments in our understanding of key attenuation processes and technologies for evaluating natural attenuation processes, and a changing institutional perspective on when and where Monitored Natural Attenuation (MNA) may be applied. Despite these advances, restoring groundwater contaminated by anthropogenic sources to allow for unrestricted use continues to be a challenge. Because of a complex mix of physical, chemical, and biological constraints associated with active in-situ cleanup technologies, there has been a long standing focus on understanding natural processes that attenuate groundwater contaminant plumes. We will build upon basic environmental science and environmental engineering principles to discover how to best implement MNA as a viable treatment for groundwater contamination plumes. Additionally we will delve into the history behind and make predictions about the new directions for this technology. Any professional working in the environmental remediation industry will benefit from this in-depth study of MNA. We will use lectures, readings, and computational exercises to enhance our understanding and implementation of MNA. At the completion of this course, students will have updated understanding and practical tools that can be applied to all possible MNA sites.

From the lesson

Modeling Tools to Support MNA

In this series of lectures, we talk about models and how they can be used to understand MNA. We first start with two key MNA questions: How Long? (will the plume get) and How Far? (how long until the site is clean). Then a review of analytical computer models, remediation timeframe models, the more complex but more powerful numerical models, and even a discussion of the “Fourth Paradigm: Big Data”.

How Long, How Far? Key Questions for Modeling Tools12:11

Fate and Transport Equations for MNA6:29

Analytical Groundwater Models for MNA11:30

Estimating Remediation Timeframes10:49

MNA Lessons from "Big Data" Studies10:05

Numerical Groundwater Computer Models and MNA9:46

Meet the Instructors

Pedro Alvarez
George R. Brown Professor of Civil and Environmental Engineering
Department of Civil and Environmental Engineering
Charles Newell
Vice President
GSI Environmental Inc.
David Adamson
Senior Environmental Engineer
GSI Environmental Inc.

CHUCK NEWELL: OK.

Well, today we're going to go slide into the world of the complex

and the sublime sort of with a discussion of numerical computer models

that can be used to evaluate MNA.

DAVE ADAMSON: Right.

So we've talked about that more complicated models where

finite-difference are finite-element codes

are used to discretize the subsurface.

Then you apply transport processes, really attenuation processes,

to the plume.

But how would you put all of this into context?

Is there a hierarchy of groundwater models?

CHUCK NEWELL: Well, if you think that the first tool was that bone that

was thrown by that monkey in that movie 2001: A Space Odyssey,

he was probably using that as a weight for the bailer, right?

But we've would come a long way.

So let's talk about the continuum of groundwater tools

in this slide right in here.

So it sort of starts out that there's different things like-- terms

on the left, we go from Limited to Site Data

to Site Data with Simplified Assumptions and then

to more complex tools and models and things like that.

DAVE ADAMSON: So the first category, you have things that are hand calculations.

So that would be like calculating the groundwater velocity

or something like that.

CHUCK NEWELL: Exactly.

And so which one-- how would you do this, though?

Would you use the Darcy velocity or the seepage velocity

if you wanted to know how long a plume is getting?

DAVE ADAMSON: I would use the seepage velocity.

CHUCK NEWELL: Very good.

Exactly.

Then going down, examples of sort of site data.

It's like different scenarios documents that we

develop where you put these different plumes in different buckets

like taxonomy.

DAVE ADAMSON: What about the analytical models, then?

CHUCK NEWELL: Some examples of those, we've talked

about-- Biochlor, REMChlor, REMFuel.

And then you slide into these numerical models,

which we'll talk about today-- things like MODFLOW, MT3DMS, Tough,

things like that.

And they give you all of this detailed site-specific stuff.

So let's first just talk about some of the strengths

and the weaknesses of these numerical models.

So what we're really looking at is that their strengths,

that they can really sort of look at this macro-scale heterogeneity,

multiple layers, changing hydraulic conductivities.

They can simulate recharge and discharge to stream

so you can really connect to the surface in a good way.

They do more complex reactions for sorption and biodegradation.

And just overall, they're more realistic.

But they do have some limitations, right, Dave?

DAVE ADAMSON: Well, obviously, these things

take a little bit more expertise in order to use them.

So there's a time investment, there's maybe a money investment

that would be associated with these.

Then as we'll discuss further, you have to be

careful with how you handle matrix diffusion in terms of these models.

And then, finally, you have to enter your source concentration versus time,

right?

CHUCK NEWELL: That's right.

You know, in those analytical models, you

can actually put a source mass in there, and it will you

use this box model to tell you what those concentrations are versus time.

These numerical models typically don't have that.

You have to actually enter in, this year, it's this concentration.

This next year is this concentration.

You may have to run something like REMChlor to get those numbers.

So one of these subtle things that doesn't have these built-in source

models.

Here's something from the RT3D, which is a numerical model.

They call this reactive model.

They have these frequently asked questions,

and they talk about, when do I use the simple tool vs. The complicated tool?

And in the middle here, they talk about the simple analytical models

can be useful for upfront screening assessment or some sites that

lack complexities in flow, geology, and reaction that have simple geology

and things like that.

But then they say that this RT3D numerical model, on the other hand,

can be used to model the complexities of the flow

and the geology and the reaction for natural attenuation

or for some sort of accelerator remediation technology.

So a good sort of way to balance this different stuff out.

DAVE ADAMSON: OK.

CHUCK NEWELL: So now let's look at what you mean.

You talked about discretization, right?

So what we're going to do, instead of just assuming this 1D flow analytical

model, we're going to break up that subsurface

into a bunch of different blocks.

And so, on the left, this is coming from one of the manuals for mod flow,

and it talks about, you can have these different elements in there.

And each one of the boxes, each one of these cubes,

is going to have a representation of what the geology is in that space.

And then you can define different cells as being boundaries.

This one's creating flow.

This one's sucking out flow.

Then you can sort of get complex-side geology on the right

by sort of changing the thickness of your cells

as you can see in that example.

DAVE ADAMSON: So you maybe have some complexity,

but you're trying to at least approximate it with these grids.

CHUCK NEWELL: Yeah.

So you can really get complex hydrogeology, different layers,

the different thicknesses of the different units

with doing stuff like this on the right.

So now let's go look at the reaction terms.

And so here's this sort of what they have in their manual.

And you start to get the sense this is a little more complicated than it

is putting in this first-order decay.

You can do that if you want to represent your attenuation of a dissolved plume

that way, but you can also have other ways

to look at some of these chemical reactions and things like sorption.

Give us some examples.

DAVE ADAMSON: Well, you can do linear sorption.

Freundlich.

You can use Langmuir.

I mean, you have all sorts of choices in terms of just that one process.

CHUCK NEWELL: Right.

And so you have to consider, there's more data

that might be required to put that stuff in there,

but you have that power to do that with these numerical models.

Now let's go to this sort of document that

talks about how to apply this RT3D model for chlorinated solvent sites.

And just to give you a sense of here's this outline

if you have these different problems that you want to solve.

I have chloroethanes only, I do this stuff on the right.

If I've got a mixture of things, I do the stuff on the left.

But it talks about which dials you turn, which processes you want to use.

But you sort of get the sense we're getting more complexity here.

So there's more power, but it does take time to sort of figure this stuff out.

Let's go through a list of attenuation reactions

that are available in this RT3D numerical model here.

The list had a bunch of different ways of doing this.

Dave, what's your favorite here?

DAVE ADAMSON: I'm a bio guy.

I like the Double Monod Model.

CHUCK NEWELL: OK.

But they have the instantaneous reaction,

which was used in the bioscreen.

But it combines this power of really being able to better

simulate the complex geology.

You can get sort of a wide variety of these reaction terms

to sort of simulate all this stuff.

Let's look at an example.

DAVE ADAMSON: OK.

CHUCK NEWELL: So let's look at a numerical model.

This is a great paper by Rasa, Doug Mackay, and colleagues.

And it's basically looking at a vertical slice in the subsurface.

There's this picture with a lot of different cells.

And the reason they're using a lot of three-meter modeling zones--

so it's not that much.

About 10 feet, right?

But they have 60 vertical layers in there,

because they're modeling matrix diffusion of oxygen

through a silt layer getting into this MTBE plume.

And so the deal is that the MTBE plume is on the left,

and this is what this picture looked like from this numerical model

from 2004 to 2006.

That MTBE plume looks like it's going away.

And why is that?

DAVE ADAMSON: Well, it's degrading, and integrating in part to TBA,

and that's shown on those right panels.

So you see the purplish color sort of increasing as you go through time.

CHUCK NEWELL: The daughter product, right?

And so this model will handle some of that reaction,

but then it also handles this very sort important piece

that you could not do in an analytical model.

It has that layer that the oxygen is diffusing through.

Now we do have this-- we talked about this in some of the previous lectures--

this matrix diffusion piece, because that's a very low permeability

that the oxygen's going through in their model.

You have to be careful with that.

And here was this paper from two of the folks

from the University of Guelph-- Steve Chapman, Beth Parker.

Tom Sale from Colorado State, Lee Ann Doner.

But they tried to model this green tank experiment

that we showed in several lectures, right, about matrix diffusion.

And let's look on the bottom left.

How many of these nodes or these cells do they

need to model this in MODFLOW MT3D?

DAVE ADAMSON: Well, they got up to 10,000 nodes.

CHUCK NEWELL: They needed a lot.

Very thin layers through this.

And what was their key conclusion right now with numerical models when

you're looking at matrix diffusion?

DAVE ADAMSON: Well, in this case, the quote's "requires much higher

resolution than commonly practiced" to simulate

this process of matrix diffusion.

CHUCK NEWELL: So just a warning out there

if you're really thinking about this matrix-diffusion process-- you

may need to be careful about how you do this to make sure you capture this

correctly.

Still a lot of work being done in this area.

But one thing you could do is, if you don't want to apply these models,

there are these things called these type sites.

And this is in this State of the Science review of management of contaminants

in low-permeability probability zones.

Dave, a little bit about that project?

Who's the authors on that?

DAVE ADAMSON: Its primarily Tom Sale, but Beth Parker, yourself, and--

CHUCK NEWELL: And you.

DAVE ADAMSON: Well, yeah.

I contributed a part of this, yeah.

CHUCK NEWELL: So this one said there's a modeling chapter where

they set up one of those really, really powerful models-- HydroGeoSphere--

and then would model what they call different type sites.

And so if you don't have the time to model something, you could say,

well, I think my site's most like this two-layer sand or clay,

or my model's got a fractured network.

Its fractured rock.

And lets look at that example here.

And you can sort of see their domain, and that's showing up on the left.

What are all the lines up there in that top panel?

DAVE ADAMSON: Those are individual fractures within that domain, right?

CHUCK NEWELL: So you might say, in Texas terms, that's a mess of fracture.

DAVE ADAMSON: That's a mess of fractures.

Exactly.

CHUCK NEWELL: So they sort of put those in there, and they say,

this is a highly fractured rock.

And then they've got different sizes of aperture size and the frequency

and hydraulic conductivity and the concentration,

and then they let it rip.

And then they can say, well, you can see these graphics in here

just by observing.

If your site sort of looks like this on the top left, maybe not exactly,

but you can get an idea of the style of the site.

And this shows that, over 20 years, 50 years, 100 years,

this is what this plume would look like.

And in this case, they've got some degradation in here,

so that's why it starts to fade out after year 100.

So some powerful stuff in terms of these type sites and that

if you're really looking at matrix-diffusion, chlorinated-solvents

type things.

So let's wrap up and just say that numerical models can model things that

analytical models cannot.

DAVE ADAMSON: And then you pointed out several of the limitations,

and they're powerful, but you don't often have a modeled source term,

and you may need those really fine grids in order to capture matrix diffusion.

CHUCK NEWELL: And the type site can be sort

of a middle ground of looking at this stuff.

You can see what a sight that sort of represents your site might look like.

Explore our Catalog

Join for free and get personalized recommendations, updates and offers.

Coursera provides universal access to the world’s best education, partnering with top universities and organizations to offer courses online.

© 2018 Coursera Inc. All rights reserved.

Download on the App Store

Get it on Google Play