You are on page 1of 5

# Tanner Phillips

Brandon Soderquist
Scott Siegel

Section 1

December 6, 2014

## Effect of Solutes on Ice Cube Melting Time

Introduction
The purpose of our experiment was to test the effect of different solutes for
an ice cube to melt. The main effects we were interested in were (a) if solutes in
water effected the time it took an ice cube to melt, and (b) if solutes in ice cubes
effected the time it took the ice cubes to melt. We were also interested to know if
there was an interaction between certain solutes. We are expecting to see that
solutes in water will cause the ice to melt slower, and that solutes in ice will cause
the ice cubes to melt faster.
The first step in our experiment was to choose what solutes we were
interested in. For water, we chose salt, as well as sugar as solutes, and also
included a control group with no solutes. For the ice cubes, we chose sugar and
acetic acid, and also included a control group with no solute. Acetic acid is the main
ingredient in vinegar other than water (the vinegar we used was a 5% acetic acid
solution), and we will refer to the acetic acid solute as simply vinegar water.
Design and Data Collection
After we had decided on the solutes we were interested in, we determined
the appropriate statistical model for our experiment. While measures were taken to
control for the temperature of the water across the entire experiment, there was a
real concern about the water temperature variation. For this reason we decided that
the best statistical model to use would be a SP[1,1]. This design allowed us to block
for each separate tub of water, and yields the following model:

## yijk = + i + j(i) + k + ()ik + ijk,

Where is the mean of our data, is the type of water used, (which is
nested in ) is the tub (plot), is the type of ice cube used, and is the residuals for
each ice cube. Both and have three levels:
: Sugar water, Salt water, Water
: Sugar ice cube, Vinegar ice cube, Normal ice cube
It was also decided that each tub would contain 1.5 cups of water, and have three
subplots (as shown above): left, right, and center.
After doing a round of preliminary testing to determine approximate
(conservative) between and within variances, we did a power analysis to determine
how many ice cubes we would need to support an 80% power level. Using
power.anova.test in R, and treating the experiment like a BF1 with 9 treatments, we
concluded that 3 ice cubes per treatment was more than sufficient (R returned an
n of 2.03). This meant we would need a total of 12 tubs for a balanced experimental
design (for a full table of the data see the appendix).

Tanner Phillips
Brandon Soderquist
Scott Siegel

Section 1

December 6, 2014

Before experimentation began both the ice cubes and water needed to be
prepared. Because water does not hold solutes well at a cool temperature, 48 hours
prior to testing the water for the sugar ice cubs was heated and salt was added at a
proportion of 1 cup per liter, and all ice cubes were measure at 1 tablespoon and
frozen. 24 hours prior to test, both salt and sugar water were warmed and solutes
were added at a rate of 1 cup per liter. All water was left over night to cool to room
temperature.
Prior to the experiment we randomly assigned each of the tubs one of the
three water types using R in case of a systematic bias such as the temperature of
the room decreasing over the course of the experiment, or the ice cubes warming
from being removed from the freezer multiple times. For each tub we randomly
selected a treatment for each ice cube using R (Left=1, Center=2, Right=3).
This was to block for potential bias caused by one side of a tub having a
different effect. We took out the ice cubes being used and each of us placed
one ice cube on the left, center, or right of the tub, ice cubes were then returned
to the freezer. Using timers, we patiently watched the ice melt until there was
nothing to be seen. One experimenter was chosen as the recorder so that there
was a consistency in deciding when an ice cube was gone. We recorded the
time and waited for all of the ice in that tub to be melted. When the tub was
cleared we rinsed the tub and then filled it with the next solute as previously
determined. We ran two tubs at a time, and placed a light over them in order to
make the ice cubes easier to see. The experiment was repeated until all ice
cubes were gone.
Data Analysis
Source

DF

SS

Mean

Water

Tub

Ice Cube

Water*Ice
cube
Interaction
Residuals

26,575,74
3.36
2,092,900.
056
468,434.9
17
844,112.8
89
54,314.94
4

Total

36

18

330,088.8
33
30,365,59
5

Corrected
SS
26,575,74
3.36
1,046,450.
028
52,048.32
4
422,056.4
4
13,578.73
6
18,338.26
9

## The results for our experiment our as followed:

Main effects

F Value

P-Value

20.11

0.0005

2.84

0.0285

23.02

<0.0001

0.74

0.5767

Tanner Phillips
Brandon Soderquist
Scott Siegel

Section 1

December 6, 2014

## Water: The main effect for water was shown to be

statistically significant, with a p-value of .05%. Because we
used a SP[1,1] design, we used the Corrected SS for Tub as
our error in this analysis. When comparing the three group
means, we saw that Tukeys HSD gave us more power than
a Benferroni comparison of means. We found that normal
water with no solutes melted water significantly faster
than water with both salt and sugar as a solute. Means are
shown in figure to the right.
Ice Cube: The main effect for ice cube was shown to
be statistically significant, with a p-value of less than .01%.
When comparing the three group means, we saw that HSD
gave us more power than a Benferroni comparison of
means for this analysis as well. We found that ice cubes
with no solutes took significantly longer to melt than either
the sugar water or vinegar ice cubes. Means are shown in
the figure to the right.
Other analysis
Blocks: As a side note, it is
worthwhile to notice that our choice of
experimental design was validated by
the effect for Tub (our plot) being
significant with a p-value of 2.85%.
While we were not able to keep biases
out of our experiment, because we
randomized at the plot and sub-plot
levels, we were able to control for the
lurking variables that may have caused
a systematic change in temperature
over the course of our 3+ hours of data
collection.
Interactions: We did not find a
significant interaction between water
type and ice cube type in our study, with
our f-test returning a p-value of 57.67%.
Because the complexity of interactions,
it is useful to refer to the interaction plot to the left. While the lines are not perfectly
parallel, there is a general trend that is evident.
Conclusion
When looking at our study overall, we come to the conclusion that the time it took
to melt ice cubes was in fact effected by the presence of solutes. We find that liquid
water without the presence of a solute(sugar or vinegar) melted the ice cubes
fastest. Solutes in the ice cubes themselves showed a different conclusion. Ice
cubes that were without solutes took the longest amount of time to melt. There was
no interaction of significance in our experiment which enables us to say that water
without the presence of a solvent melts all ice cubes tested the quickest, and that
the cubes without a solvent were the slowest to melt in any of the solutions tested.

Tanner Phillips
Brandon Soderquist
Scott Siegel

Section 1

December 6, 2014

## We designed the experiment as a SP[1,1] which enabled us to control lurking

variables. This control as well as randomization of the treatments at all levels
allows us conclude that the relationship between solvents and the time it takes for
the ice to melt is causational. The population is limited to those few solvents that
were tested. If the experiment were reproduced with more time and resources it
would be curious to see the results of a different solutes and their effects on melting
time both in the melting solution as well as in the ice cubes themselves, also while
no significant interaction was found, the most suspicious point seems to be where
sugar ice cubes were melted in sugar water, and if further studies were to be done,
it would be worthwhile to see if there was an interaction between sugar water and
sugar ice cubes, or if perhaps there was even between any ice cube with the same
solvent as the water it was dissolved in. While the scope of our experiment was
limited to a few solvents, the results were clear and causational.
Appendix
*reads in data file;
data ice;
infile
"C:\Users\Tanner\Documents\data\icedata.txt"
delimiter=" " firstobs=2;
input Plot Water \$ Spot Icecube \$ Time;
group = 10*icecube + water;
run;

## power.anova.test(groups = 12, n = NULL,

between.var = 1300, within.var = 900,
sig.level = 0.05, power = .8)

## *Prints data file;

proc print; run;
*performs analysis of data using sp[1,1]
design and
uses tukey HSD to compare all means for water
groups and ice groups;
proc glm data=ice;
class plot water icecube;
model time = water plot(water) Icecube
Icecube*water;
test h=water e=plot(water);
means water/ tukey;
means icecube/ tukey;
run;
*show us the means of the 9 different
'treatments',
used to create interaction plot by hand;
Proc sort data=ice;
by water icecube;
run;
Proc print; run;
Proc means data=ice;
by water icecube;
var time;
run;
*used to create base for interaction plot.
Chose to compplete plot by hand
as output was not particulary useful;
proc gplot data=ice;
plot time*water=icecube;
plot time*icecube=water;
run;
#used to determine number of data points needed in
each treatment

Plot

Water

Spot

Ice cube

Time in
Seconds

normal

sugar

387

normal

vinigar

485

normal

water

685

sugar

sugar

730

sugar

vinigar

630

sugar

water

normal

vinigar

595

normal

sugar

325

normal

water

774

sugar

sugar

795

sugar

vinigar

1095

sugar

water

1235

salt

water

675

salt

sugar

850

salt

vinigar

790

salt

water

salt

vinigar

1245

1470
985

Tanner Phillips
Brandon Soderquist
Scott Siegel

Section 1

salt

sugar

salt

vinigar

salt

water

salt

sugar

water

water

water

sugar

sugar

sugar

December 6, 2014

1085

10

water

vinigar

430

960

10

water

water

625

1290

10

water

sugar

410

655

11

sugar

water

1345

vinigar

390

11

sugar

sugar

sugar

405

11

sugar

vinigar

1130

water

725

12

salt

sugar

1030

water

1470

12

salt

vinigar

sugar

960

12

salt

water

vinigar

1175

865

940
1290