Today we had our final summer research event: oral and poster presentations.  The students did a wonderful job presenting their work.  I was very proud of them.  Here's a picture of the students with the poster. Download the poster!

Today we ran another layer set up. This set up was the same set up we did before with 6 inches of lake superior below 3 inches of F65. we were running this setup again because the experimental data did not match up with what the model predicted. looking at the results from todays trials, the layers acted in a similar way to what they did before. For monday, we set up the small cylinder with lake superior sand to run again. the data for the 3 different size cylinders was not quite what we were expecting. we are running this cylinder again to see if the results will be the same. We will feel more confident with the data if the results are the same as before. We also set up the medium cylinder with lake superior sand as well. In the medium cylinder set up, we placed the column in and added sand to the outside of the cylinder only. we will add sand to the column during the trials to see the effects the difference between the inside and outside levels of sand have on the k value. We have had situations where the sand level is higher inside than outside, as well as the sand inside being lower than the outside.

Tomorrow we plan to do some more work on the poster and perhaps start working on the formal presentation we will have to give in a few weeks.

Group Picture

Fake fire alarms at lunch prompted group photo shoot, here's the entire group!

Today we tested the Superior sand in an even larger cylinder than before. We used our method of the falling head test and after three runs, we would remove the column and wait 5 minutes before running more tests. This was done to make sure that the sand had time to settle. We have not analyzed this data yet, but hopefully we will be able to make a comparison of the small, medium, and large cylinders. Tomorrow we will be running a layered set up with the Superior sand on the bottom and F-65 sand on top. We did not obtain the best of data last time with this set up so we are testing again to hopefully see a better set of data this next time.

Menke Observatory Trip

Relaxing, while the rest are still arriving...

Hiking

 Throwing bean bags.....

Dr. Mitchell shows us the telescopes.

Moon taken through telescope. (Saturn was my favorite, but no picture.)

An Update

This weekend we went to a River Bandits game. We all enjoyed the game, even if the Bandits did get beat up pretty bad, as it was Rascal's birthday. Tonight we will be going to a star party where we plan on having a cookout and enjoying the day before observing the stars.

As our time for research is beginning to come to an end, we are focusing more on analyzing results and running tests that will enhance what we have already learned, or will indicate causes of error. In the lab we have been trying to narrow down what setups would be best to run that we believe will provide interesting and useful results based on all the tests we have done up to this point. Today we learned that putting marbles into the sand isn't having a consistent effect, if it is having an effect at all, on our calculated K values. With this knowledge we are turning our focus back to setups with different layers of sand and we are planning to run setups similar to ones we have run before to confirm results. Another important test condition we will be testing, so that future in lab falling head tests can be run more effectively, is that of the effect of the size of the outer glass container. Our models created through MODFLOW show that the size of the system has an effect on the K value that you measure, but we want to determine if this is a true noticeable effect in the lab.

Pictures don't require words....

River Bandits Baseball Game

3 layers of sand...

 Adding Lake Superior sand...

We ran tests for the superior sand in the smaller "hobby lobby" cylinder and got experimental results very similar to our experimental results from the larger "Michael's" cylinder.  We had hypothesized that there would be a significant effect from the boundaries due to some preliminary modeling of the smaller tube.  The results are not so far different than the modeled ones that we can definitively say that the model is over representing edge effects, but it seems that way.  Below is a graph of our data.  The first is how we were looking at it.  The second is the same, but I changed the axis, because I was finding the top graph to be rather sad, but really our data is all very close together, which you can see when 0 is included, so while we are struggling to know what to think of the modeling vs. experimental results for a single sand, this is only a problem because the model is only off a little from the experimental, so distinguishing the sources of the problem is challenging.  (and this sand is quite significantly different from the 65.)

Pure Superior Sand

changed axis

The first setup that we tested was the one with a layer of 100 marbles about 3 inches above the bottom of the outer column with about 6 additional inches of sand on top of the marbles. When testing with the falling head test, the column was always above the level of the marbles. We have previously done this test with the f65 sand and today we were testing the lake superior sand. After finding one of our columns broken yesterday, Susa went out and bought 2 new columns. One is the same as the the columns that we have had before the other was a bit taller and thinner than the columns that we have been using. The two new columns were set up yesterday afternoon, so that would be allowed to settle overnight like we have been doing. The big column was prepared with lake superior sand and only 7 marbles placed inside the column. There was not a huge change in the k values when only 7 marbles were placed in the column and no marbles were outside. The smaller tube was filled with some f50 sand for further testing since we have not used that sand since we changed our method. Since this column is much thinner we think that there might be some edge effects from the two columns used in the set up of the falling head test being so close together. To solve this problem we placed some superior sand in the thinner column and will test it tomorrow. We chose superior sand since we feel that we have some really good data on this sand to compare to the k values that we will get tomorrow to see if the thinner column effects the k values. Tomorrow we also plan on lodging some more extensive testing of the superior and f65 sands in terms of distance from the absolute bottom of the column and the k value trends.

MODFLOW modeling vs Experimental data

The first test that we ran using marbles was the of a layer of 100 marbles below the column in F-65 sand. We ran this test twice a weekend apart since we suspect that the rate in which this particular sand settles is rather slow, and that different degrees of settledness effect our measurements. Below is a graph that depicts the two times the sand was run (trial 1 and trial 2) along with 3 different models, of what we might expect the K values to have been, from MODFLOW.

 
The model with marbles was meant to replicate the conditions of our experiment exactly. The model with a raised boundary is a comparison of having a layer of marbles with sand below or having the bottom of the glass container right below the column. The model without marbles is just the F-65 sand run through MODFLOW with no obstructions.

The first thing we noticed is that the two times we ran this experiment we got graphs with similar shapes, but allowing the sand to settle brought the average K value down quite a bit. The models indicated that there is very little effect on the measured K value unless you are right above the layer of marbles or glass bottom (boundary) of the container. The second time we ran this experiment was more similar to the models than the first time as it shows a drop right where the models showed a drop at 0.15 meters (right above the marble layer).

The next test we ran once with both sands (F-65 and Lake Superior). This experiment involved a layer of marbles in which the column was going through. There were a total of 97 marbles in this layer, 7 of which we had placed within the column. Below is a depiction of our experimental data vs our models at this point.


The two models show the same shape and are at different K values as the two sands have different K values. The accuracy of our model is in question and we are working to develop the most accurate one we can. The two experimental runs showed some what different shapes, but both did indicate an initial drop in K value, a leveling off and an increased value at approximately 0.13meters (more noticeable in the Lake Superior sand).

We plan on running some more different marble set ups as well as perfecting our means of modeling what should happen.

Yesterday and today we have been busy running tests that involve the use of marbles acting as rocks in the sand. We ran a test with 90 marbles outside the column and 7 marbles inside the column. This test was ran using the Lake Superior sand. The marbles were placed in 6 inches of sand then the marbles and then another 3 inches of Superior sand above the layer of marbles. Another test we ran dealt with the Duck Creek sand. We finally got to test it after the mishap of the glassware breaking. David will be blogging later talking about how these tests compare with what he is running through Matlab. Today when we came into the lab, another large glass container had cracked sometime over night. Due to this we were unable to run the setup that was in the glass container. Tomorrow we will be testing the hydraulic conductivity in the Superior sand when the marbles are at 3 inches in the sand and the column is always above that layer.

Monday....A Little late

Today we ran the set up with a layer of 100 marbles below the column another time. The times this time were slower than the trial we did the last time we met, but I am not sure what this did to our k-values, and I haven't compared the data between the two trials. The F-65 sand that we are using seems to take a long time to settle, so it is possible that when we ran the falling head test with this set-up before the sand was not fully settled. We let the sand sit for the weekend before running the test again so the sand should have been settled this time. We got another column, replacing the one that broke when we were going to test the Duck Creek sand last week. Because we thought that there is a problem with the F-65 sand and settling, we are planning on letting each set up settle overnight. Yesterday we prepped another marble layer test. this time the marble layer is 3" deep. the marble layer is present inside the column and outside the column. We also prepped the other container with some of the Duck Creek sand in an attempt to actually measure the k-values this time. The Duck Creek sand causes the water to be very cloudy and murky, so we are allowing this sand to set up overnight in an effort to let the water clear up so we can actually see the column inside to read the measurements.

I've got a code working for placing 100 (or the number you specify) circles inside the bigger circle.  Here's an example image of the placement.  I think it looks pretty realistic.  It places 100 inside at random and then adjusts each at random until none overlap by calculating the amount of overlap each time and only moving if the overlap decreases.  It's not in your matlab-to-modflow code yet, but it's a step in that direction.

Example generated image of 100 marbles in a layer

Movie done in NetLogo showing a very similar algorithm to the one we used in MATLAB.  Red marbles are overlapping and blue ones are not. 

This morning when we got to the lab, the cylinder that contained the Duck Creek sand had broken sometime during the night. Due to this unfortunate event we were unable to test that sand today. We did however have another cylinder set up that contained 6 inches of Lake Superior sand on the bottom and 3 inches of F65 sand on top of that. We used the falling head model to obtain our values and calculated our k-values. We then compared the experimental k-values with the k-values we got using Modflow.  Although they were not the exact same, the general trend was there. When the F65 and Lake Superior sand were flipped, the graph had a trend going the other way however the experimental results seemed to more accurately match the k-values obtained from Modflow when the Lake Superior sand was on top. We are not exactly sure why, but David put forth that it might be because the Lake Superior sand seems to settle much better than the F65 sand.

After lunch, we also ran some more tests. This time we set up a single layer of rocks that covered the sand. We then put 6 inches of sand on top of the layer of marbles. We wanted it to try to mimic a layer of rocks in a stream. We started the column a half inch above the rocks and continued to move the column up obtaining our values need to calculate hydraulic conductivity. We will look to compare the k-values we got once we have obtained k-values from running Modflow.

broken cylinder with Duck Creek sand

adding the marble layer

100 marbles in layer

Duck Creek Adventure part 2

Today was an early start to survey a new location at the creek. As before we used the falling head test at multiple locations in the creek. This section of the creek turned out to contain sediment that was more rocky in nature than the more sandy conditions at the first location. There was a top layer of sand that was rather shallow with a layer of rocks underneath. There seemed to be this layer of rocks at approximately the same depth across the whole area tested, so we expect differences in the values calculated for this location compared to the first. We collected sand from the location we had tested to bring back to the lab so we could try to confirm the K values by testing the sand in the lab. We set the column up to be tested tomorrow as the water was very murky and we wanted to allow the sand time to settle.

Once we were back in the lab we proceeded to test a new setup of two layers of different sand after having let it settle overnight. We tested different depths into the two layers of sand by starting with the column deep through both sands and slowly moving the column up half an inch between trials. Our goal from setting the test up and while carrying it out was to disturb the layers as little as possible to be more sure that the interface between the layers was where we thought it was and that the layers didn't get mixed greatly with each other. A depiction of the comparison of our measured experimental values with our model values generated with MODFLOW is shown below.



Below are some images from our testing the new location at the creek and of our layered sand test.

Matt and David

Matt and David (with Jessica behind on bridge)

Superior sand on top, F65 on the bottom

Today we finished up the falling head testing of the Lake Superior sand. We now have 10 trials for the Lake Superior sand and the f-65 sands, which are the 2 sands we will be focusing on when we start with the layering of the sands. We wanted to choose the 2 sands that had distinctly different k-values which would be the F-50 and the F-65, but these 2 sands are the same color. The Lake Superior sand is a much darker color and has a k-value that is similar to the F-50 sand so we chose to use it to tell the layers apart better.  Before we called it a day we set up the first of our layer tests. The first layer test will have the F-65 sand on the bottom and the Lake Superior sand on the top. The column was placed into the first sand and then the second sand was added to the inside and outside of the column. The sand will be allowed to settle until we get to testing it tomorrow afternoon or thursday.
We also experimented with finding the porosity of the 2 sands. This was done by filling a graduated cylinder with a known amount of water and then pouring some sand into the column. Once the sand was allowed to settle, the level of the sand and the level of the water were measured to calculate the porosity.
Tomorrow we plan on going back out to Duck Creek to do some field work. I think we are going to go to a different place along the creek and will be doing measurements that are similar to what we did the last time we were there to compare the K-values at different parts of the stream.

F65 and Lake Superior sand

Week 3

Monday was another very productive day. We completed testing on the Lake Superior sand and F-50 sand using the constant head method, however those k-values obtained were not all that similar with the k-values obtained using the falling head method. We felt that the constant head method, had a larger chance of error, so we are going to proceed with the falling head method.

Due to this decision, in the afternoon, we ran more tests on the Lake Superior sand to get a total of 10 k-values and average them. We decided on this because the more trials we have, the better our statistics will turn out.  Each time we placed the column into the sand we would record the level of how far the column was in the sand, the height of the water, the initial and final heights the water drops, and the time. We would run three tests without removing the column and take the average of those three tests. Then we would take the column out, wait 5 minutes for the sand to settle and repeat that procedure until we had 10 k-values. After testing the Lake Superior sand with that method, we then did the same for the F-65.

While Jessica and I were performing those experiments, David was experiencing the joys of Matlab by testing what would happen when we have two layers of sand. From the computer simulations, we will soon be trying out what happens in the lab setting when we put two layers of various sands in the cylinder. Hopefully what we find out that the lab will be similar to the computer simulations on Matlab.

The Constant Head Test

After days of working to produce a secure, water tight, and effective constant head column apparatus, testing using the method began today. The difference with this test compared to the falling head test that we have been using up to this point, is that the water level in the column in the sand is meant to stay constant. Water is constantly pouring into the column, and excess water is caught as the water level in the column is kept constant. By knowing the amount of water added to the column and the amount of excess water that was collected we are able to determine the amount of water that went through the sand. This allows us to determine the hydraulic conductivity of the sand. 
A diagram of the setup is displayed below.

Diagram by Susa

We have been working to develop an effective method to add water from a separate cylinder while maintaining a constant water height. Right now the cylinder is prone to add water much faster than our spout exiting the column loses water. To maintain a constant water height we have been starting each trial holding the cylinder low and moving the cylinder up as water drains out to maintain a constant water height. Of course there is always variation in what the water height actually is which adds a level of error to the method. By the end of the day we had developed a method that was working moderately well for us, but we are still considering possible changes that would reduce error.


The reason behind using a more complicated test to determine values we can determine using a simpler test is that when using the computer and MATLAB to perform simulations being able to input a single water height is much simpler. We plan to continue improving our constant head test techniques as we test all the sands we have tested with the falling head test. If we find that both tests produce the same K values then we will continue to use just the falling head test in the lab while simulating constant head tests using MATLAB.

Constant head column (design 2)

The tube seems really sturdy now.  It's drying.  Hope it's waterproof, I smothered one side with the silicon glue, but couldn't easily get glue into the inside.















 It wasn't, here's Jessica applying more silicon (6/5/12)...

.

Another Tuesday

Today we tested the last of the purchased sands using the falling head test. The next step in our project is to get the apparatus set up for the constant head test. We went to set it up today, but as we went to attach the required hose the silicon glue that was used to make the attachment water tight broke. To solve this problem we took a group trip to Menards to try to find other pieces that would work better then what we had. We ended up getting some new hose attachment pieces that actually screw together so we don't have to rely on the silicon glue to hold it together. The glue will still be used to make sure all the connections are as water tight as possible. The new pieces are larger than the piece we had before so the hole in the tube has to me made larger, which is a difficult process with the limited tools we have present in the room we are working in. Tomorrow is the 4th of July so we wont be meeting, but Thursday morning we plan on meeting in the Library to do some preliminary work on our poster while Susa makes her way back to school from the Chicago area. The plan is to do some work with the constant head test Thursday afternoon if the new pieces are attached properly by then.

Duck Creek Cross Section K values

Posting here because I don't know if you can reply to a comment with an image. Plotting the K values as you move across the creek from one side to the other gets you this chart.





















The K values look somewhat random to me, if anything they might increase as you moved across the creek with a couple outliers.

Duck Creek Adventure 1 Pictures

Matt, Jessica, and David arriving with field supplies (Galvin Park on Duck Creek)

Susa and Matt filling column

David, Matt, Susa, and Jessica

Jessica (recording) and David (timing)

Matt and Susa (making measurements)

David, Susa, Jessica, and Matt after the measurements....

Week Two

We decided that we would start this week off by going out into the field  to do some research. We went to Duck Creek, which is a small stream close to St. Ambrose. Even though we got out there at roughly 9:15AM, the heat and humidity were quite high. I decided to wear waders so I wouldn't get wet. Into the water Susa and I went, while David and Jessica stayed on the side of the stream. We decided that for our first time out running field tests we would just try and see if the falling head setup would work. The first trial was successful and from that position we decided to move up the stream about two or three steps each time. At each position, we would run three trails in order to obtain an average of the data. We then decided to test a cross section of the stream. In doing so we would run three trails at a position and move half a meter horizontally across the stream. We continued to do this until we got to the opposite side. The last meter we hit mud when putting the column into the stream. Due to this the test didn't work and we couldn't calculate any K-values. The area of Duck Creek that was chosen seemed to work well and hopefully the data obtained will be able to guide us in the right direction. As for the waders, well they weren't so successful. There ended up being small cracks in the them and by the end of testing, I had standing water in the waders. Next time, I don't think I will be wearing them. For tomorrow we will be running more tests back in the lab setting, continuing our testing of the different sands that we have.

Not all work, but a play...

Back Row: David, Jessica, Ashley, Katie, Jamie, and Dr. Trujillo                                                                                            Front Row: two people we didn't know, then Jen, Alicia, Kevin, Aubrey, Marie, and Dr. Stonedahl (Photo credits Forrest)

We saw the Shakespeare play Measure for Measure tonight in Rock Island.  

Earlier in the week we played some frisbee and hit around a volleyball in very hot weather.

                                       Back row: Jamie, Dr. Trujillo, Marie, Matt, Jessica, David,  Dr. Stonedahl                                   Front row: Alicia, Jen, Aubrey (Photo credits: Forrest)

End of Week 1

Week 1 of the summer research institute has come and gone. At the end of this first week we seem to have a pretty good grasp on the running of the falling head test along with how altering the different parameters effects the resulting k value.  Today we decided to switch things up by adding some marbles into the sand to act like rocks. We started with just a few marbles layered on the sand topped with an additional layer of sand and worked our way up to over 50 marbles. The falling head test was run just as before by placing the smaller tube into the sand and filling it with water and timing how long it takes the water to fall a certain distance. The addition of the marbles took longer for the to fall making our calculated k value smaller as expected.

We also started to look and try to comprehend the Matlab code that we will actually be using to produce and image with our data. The code is much different than what we were using before to practice and seems to be a little confusing at the moment, but I am confident that we will be able to get in figured out in the next five weeks.

Next week we will begin our exploration into the constant head test, starting with the construction of the required apparatus. We can't fully start the exploration until we obtain a new glass cylinder. The glass cylinder that was being used became the first casualty of our research, shattering while Matt attempted to remove the sand and marbles. Next week we will probably head over to Duck Creek to do some preliminary research into the sand/soil composition, seeing if the sand/soil will even allow us to take measurements and maybe get some tests done to see what potential problems we might come across. 

Column with marbles (picture added by Susa)

More testing, more data

Another day of testing of the F-60 sand and more of the parameters have been acquired. We figured out that the reason for the differences in the k-values from the previous day was because the sand in the inner column was higher than the sand around the inner column. We tested this and found that by adding more sand into the inner column it changed the k-value. From now on we will figure out our length by how much of the inner column is in the sand by using the inside amount. This change helped us to get more accurate k-values. We then moved on and started running a few tests with the F-65 sand. This sand is consider to be more fine which will mean there will be less room for the water to move through. Due to this, we are predicting that the k-values will be lower than that of the F-60. So far in the few tests we have ran, the data looks that way. Whether there will be a statistical significance is yet to be determined!

Yesterday we had consistent results until we stirred up the sand and reinserted the column of sediment.  Then the K value dropped, we hypothesized that this might be due to settling of the sand.  We tried leaving it set-up and then this morning we tried it again, but the faster K values remained....  We're now hypothesizing that the system is very sensitive to sand levels inside and outside of the column (supposed to be the same, but don't seem to end up that way.)  We're trying it again with different levels of sand inside the tube, which we think is the more important value and hope that using the inner sand level will stabilize values...

The many uses of MATLAB

As we begin our research with experiments in the lab, we are also learning a whole new language on the computer. Learning the MATLAB language is a slow process, but one that should yield some useful computer simulations to compare with experiments done in the lab. To learn this coding language we jumped right in and attempted to reproduce diagrams using the code and a lot of trial and error. We started with some simple figures to reproduce and every time we would load the next image we had to reproduce we were sure we couldn't do it. Eventually though we would manage to get something right on the figure and everything else would start falling in place.

After reproducing one of the many figures I couldn't help but see the potential in the shape that I had produced. The shape was a solid rectangle with a triangle on top, and in my mind I couldn't stop myself from seeing a house. Feeling confident enough in my ability to use what little I had learned of this coding language I made it my mission to turn this figure into a full out piece of art using the MATLAB coding language.

I started with simply making a region of the diagram display as green while some displayed blue. In this I had created a lawn to accompany the house and sky. I couldn't stop there though, I then continued to code in a chimney to sit atop the roof of my house. With help from my other group members we developed a design for the front of the house and in time created a door and windows to brighten up the house. After taking a moment to admire the structure we all decided that the sky seemed very empty and that the sun would have to be created. Learning how to create circles along the way and determining how to most effectively use the code, a sun and bright rays of light were placed upon the figure.

While learning and taking command of MATLAB I managed to produce a solid piece of art. This is only the start to what I'm sure we will all learn is one of the many uses of MATLAB.