Mathematical Exploration

Modelling the Heating Curve

Personal Code:
Session Number:
Subject: Mathematics HL
––
Page Count: 14
Mathematical Exploration: Modelling the Heating Curve

Introduction

Often I make hot milk for myself when my parents are late from work. To prepare the milk,

one must turn on the stove and wait for it to reach a boil and turn off the stove before it

overflows. This usually takes a long time and for me this was a problem as I had to waste

time standing by the stove waiting for the milk to boil. Another problem was the milk never

took the same amount of time to reach a boil on different days as the ambient temperature

changes hence it was also not possible to set a timer. Hence, I decided to create a

mathematical model to accurately predict the time it takes for milk to reach a boil based on

the ambient room temperature which changes all the time.

Data Collection

In order to accurately predict the time it takes for milk to reach a boil I first needed to

collect data about how the temperature of milk changes with time on the stove. In order to

collect data I brought my small electric stove and sauce pan to school as I was not allowed

to borrow the equipment for home. After learning how to do it, I used Vernier’s1 ‘Stainless

Steel Temperature Probe’ along with the ‘Logger Pro’ software on my computer to take a

reading of the temperature of the milk every 10 second as it heated on the electric stove

until it reached a boil and started overflowing. The raw data contained 181 reading but I

have trimmed the data to only 36 point by only showing every 5th reading for the sake of

simplification. The data has been presented below in table 1.

1
1. Vernier Software & Technology. Logger Pro 3 Quick Reference Guide.
http://www2.vernier.com/manuals/LP3QuickRefManual.pdf. Accessed 4 July 2018.

| Page 1 of 15 |
Mathematical Exploration: Modelling the Heating Curve

Table 1 – Recorded Data

Time (seconds) Temperature (°C)
0 22.7
50 26.1
100 30.7
150 34.9
200 38.8
250 42.7
300 46.8
350 50.4
400 54.3
450 57.6
500 61.0
550 64.2
600 66.9
650 70.4
700 72.9
750 75.5
800 77.8
850 79.9
900 82.5
950 84.1
1000 86.0
1050 87.9
1100 89.4
1150 91.0
1200 92.4
1250 93.9
1300 94.8
1350 95.7
1400 96.4
1450 97.0
1500 97.3
1550 97.8
1600 98.1
1650 98.1
1700 98.1
1750 98.1

| Page 2 of 15 |
Mathematical Exploration: Modelling the Heating Curve

Method 1 – Digital Curve Fitting

After obtaining these values I used Google Sheets to plot a graph of the data. The scatter

plot with blue points represents how the temperature of the milk changes over time on

active stove.

Figure 1 – actual data graph

After plotting the data I notice that the graph looks like half part of a parabola hence its

mathematical representation would be a quadratic function 𝑎𝑥 ! + 𝑏𝑥 + 𝑐. I also noticed

how the y-intercept represents the ambient room temperature and wondered if it would be

possible to use this data in order to model a function which could accurately predict the

time required for milk to reach the boiling point which is 98.1 °C at any given initial

temperature (ambient room temperature).

| Page 3 of 15 |
Mathematical Exploration: Modelling the Heating Curve

I researched how I would be able to use the data in order to model an equation and found

out that this method is called ‘curve fitting’2 and it is possible to use ‘Google Sheets’3 to find

an equation for a given set of data. After watching some tutorials on how to use this tool I

was able to obtain a function and plot a graph (referring to figure 2) where the blue dots

represent the actual data and the grey line represents the function 𝑓(𝑥) given below.

𝑓(𝑥) = −0.0000277𝑥 ! + 0.092𝑥 + 21.8

Figure 2 – actual data and 𝑓(𝑥) graph

2
2. Sandra Lach Arlinghaus. 1994. CRC Press – Practical Handbook of Curve Fitting.
3
3. Patricia Goodman. 2 Dec 2014. Add a trendline to a chart in Google Sheets.
https://service.goshen.edu/support/index.php?/default_import/Knowledgebase/Article/View/974/0/
add-a-trendline-to-a-chart-in-google-sheets. Accessed 10 Jul 2018. Accessed 5 July 18.

| Page 4 of 15 |
Mathematical Exploration: Modelling the Heating Curve

The functions fits the actual data with very little uncertainty hence the function can be used

to estimate the amount of time required for milk to reach a boil. As my aim is to estimate

time with a given temperature, I needed to find an inverse function.

𝑓(𝑥) = −0.0000277𝑥 ! + 0.092𝑥 + 21.8

equating to 𝑦

𝑦 = −0.0000277𝑥 ! + 0.092𝑥 + 21.8

Dividing the equation by the coefficient of 𝑥 !

𝑦 0.092 21.8
= 𝑥! + 𝑥+
−0.0000277 −0.0000277 −0.0000277

Adding the square of the half of the coefficient of 𝑥 for making a perfect square

!
𝑦 !
0.092 21.8 0.092
=𝑥 + 𝑥+ +3 6
−0.0000277 −0.0000277 −0.0000277 −0.0000554
!
0.092
−3 6
−0.0000554

Completing the square

! !
𝑦 0.092 21.8 0.092
= 7𝑥 + 3 68 + −3 6
−0.0000277 −0.0000554 −0.0000277 −0.0000554

Squaring the fraction

!
𝑦 0.092 21.8 0.008464
= 7𝑥 + 3 68 + −
−0.0000277 −0.0000554 −0.0000277 0.00000000306916

Adding the fractions

!
𝑦 0.092
= 7𝑥 + 3 68 − 3544761.4331
−0.0000277 −0.0000554

Adding the constant to both sides

!
𝑦 0.092
+ 3544761.4331 = 7𝑥 + 3 68
−0.0000277 −0.0000554

| Page 5 of 15 |
Mathematical Exploration: Modelling the Heating Curve

Taking the square root of both sides

𝑦 0.092
±< + 3544761.4331 = 𝑥 + 3 6
−0.0000277 −0.0000554

Isolating 𝑥

𝑦 0.092
±< + 3544761.4331 − 3 6=𝑥
−0.0000277 −0.0000554

Writing the function in terms of 𝑥 and simplifying polarities. Positive values of square root

were also removed as they are outside the domain of the actual data. Giving the function

𝑓 "# (𝑥) below.

𝑥 0.092
𝑓 "# (𝑥) = −< + 3544761.4331 + 3 6
−0.0000277 0.0000554

Upon graphing the function in Desmos4, it was confirmed that this is an accurate inverse

function as it reflects along the line 𝑦 = 𝑥.

Legend

––– 𝑓(𝑥)

––– 𝑓 "# (𝑥)

----- 𝑦 = 𝑥

Figure 3 – 𝑓(𝑥) and 𝑓 "# (𝑥) graph

4
4. Desmos. 20 Jul 2018. Functions. https://support.desmos.com/hc/en-us/articles/207316093-
Functions. Accessed 21 Jul 18.

| Page 6 of 15 |
Mathematical Exploration: Modelling the Heating Curve

In order to make the function valid for any value of initial temperature, I subtracted

constant c from 𝑥 which would represent the difference in ambient temperature compared

to the initial temperature in my data which was 22.7 °C. For example, if the room

temperature is 35.0 °C, I would subtract 35.0 − 22.7 = 12.3 to give me the value of c which

would be substituted in the below function to translate the entire function horizontally.

(𝑥 − 𝑐) 0.092
𝑓 "# (𝑥) = −> + 3544761.4331 + 3 6
−0.0000277 0.0000554

Hence, using the above function it would be possible to estimate the time required for the

milk to reach a certain temperature at any given initial temperature.

Method 2 – Newtons Cooling Law

Newtons law of cooling states that the rate of cooling or heating is proportional to the

difference between the temperature of an object and the ambient temperature.5 In this

case, the highest temperature that the milk will reach can be considered the ambient

temperature forming the below equation where upper case ‘T’ represents the temperature

and the lower case ‘t’ represents the time. With the help of a video tutorial6, I was able to

form and manipulate the below equation.

𝑑𝑇
= 𝑘(𝑇 − 98.1)
𝑑𝑡

5
5. The University of British Columbia. Other differential equations.
http://www.ugrad.math.ubc.ca/coursedoc/math100/notes/diffeqs/cool.html. Accessed 4 Aug 2018.
6
6. Phil Clark. 12 Jan 2016. Differential Equations Newton's Law of Heating Problem.
https://www.youtube.com/watch?v=Kqy1xEUnEIE. Accessed 10 Jul 2018.

| Page 7 of 15 |
Mathematical Exploration: Modelling the Heating Curve

Both sides divided by (𝑇 − 98.1) and multiplied by 𝑑𝑡

𝑑𝑇
= 𝑘𝑑𝑡
𝑇 − 98.1

This is a differential equation in the variable separable format. Hence, to solve:

Both sides integrated

𝑑𝑇
C = C 𝑘𝑑𝑡
𝑇 − 98.1

New function formed after integration

ln(𝑇 − 98.1) = 𝑘𝑡 + 𝐶#

Both sides written as powers of 𝑒

e$%(' " )*.#) = 𝑒 -./0!

Power of e removed and new constant ‘a’ introduced

𝑇 − 98.1 = 𝑎𝑒 -.

Adding 98.1 to LHS and RHS

𝑇 = 98.1 + 𝑎𝑒 -.

Finding the value of a using the value of T at t = 0 from data.

At t = 0, T = 22.7 as per data

22.7 = 98.1 + 𝑎

Subtracting 98.1 from both sides

𝑎 = −75.4

Forming new equation with the value of a

𝑇 = 98.1 − 75.4𝑒 -.

Subtracting 98.1 from both sides

𝑇 − 98.1 = −75.4𝑒 -.

| Page 8 of 15 |
Mathematical Exploration: Modelling the Heating Curve

Dividing both sides by 75.4

𝑇 − 98.1
= 𝑒 -.
−75.4

Finding the natural log of both sides

𝑇 − 98.1
ln = 𝑘𝑡
−75.4

Dividing both sides by k

𝑇 − 98.1
ln −75.4
𝑘=
𝑡

In order to obtain an accurate value of k, I decided to use several values of temperature (T)

and time (t) from my data. I used Google Sheets to use the above equation for k to calculate

the value of k for all of the data. The average of these values was later calculated to be used

as the final k.

Table 2 – value of k for every 4th data point

Time (t) Temperature (T) Value of k

0 22.7 N/A
200 38.8 -0.001200989845
400 54.3 -0.001357933644
600 66.9 -0.001357933644
800 77.8 -0.001640232986
1000 86.0 -0.001829601822
1200 92.4 -0.002151950917
1400 96.4 -0.002708699303
1600 98.1 N/A
Average -0.001749620309

The value of k was also rounded off to 5 decimal places

𝑇 = 98.1 − 75.4𝑒 "1.11#234

| Page 9 of 15 |
Mathematical Exploration: Modelling the Heating Curve

Giving the final function 𝑔(𝑥)

𝑔(𝑥) = 98.1 − 75.4𝑒 "1.11#234

Just as I did with the previous method, in order to obtain values of time with temperature, I

would need to find an inverse function.

Function made equal to y in order to calculate inverse

𝑦 = 98.1 − 75.4𝑒 "1.11#234

98.1 subtracted from both sides

𝑦 − 98.1 = −75.4𝑒 "1.11#234

Both sides divided by −75.4

𝑦 − 98.1
= 𝑒 "1.11#234
−75.4

Natural log of both sides taken

𝑦 − 98.1
ln = −0.00175𝑥
−75.4

Both sides divided by −0.00175

𝑦 − 98.1
ln −75.4
𝑥=
−0.00175

Final function written in terms of 𝑥

𝑥 − 98.1
ln −75.4
𝑔"# (𝑥) =
−0.00175

Again, just like the previous equation, constant c was subtracted from the equation which

would represent the difference of room temperature from 22.7, the actual room

| Page 10 of 15 |
Mathematical Exploration: Modelling the Heating Curve

temperature from the data used to model the function. This makes the function valid to

calculate the time required to reach a boil at any given value of room temperature.

(𝑥 − 𝑐) − 98.1
ln −75.4
"# (𝑥)
𝑔 =
−0.00175

Evaluation

Using the above 2 methods I have modelled the data in 2 different ways. Below, Table 3 and

the graph in figure 4 compares the data against the actual collected set of data.

Table 3 – modelled functions against actual data and mean difference

Temp Time 𝑓 "# (𝑥) 𝑔"# (𝑥)

Data Value Difference Value Difference
22.7 0 9.81 9.81 0 0
26.1 50 47.42 2.58 26.37 23.63
30.7 100 99.73 0.27 64.09 35.91
34.9 150 149.08 0.92 100.86 49.14
38.8 200 196.4 3.6 137.26 62.74
42.7 250 245.29 4.71 176.13 73.87
46.8 300 298.58 1.42 220.07 79.93
50.4 350 347.16 2.84 261.64 88.36
54.3 400 401.89 1.89 310.38 89.62
57.6 450 450.14 0.14 355.15 94.85
61.0 500 501.95 1.95 405.25 94.75
64.2 550 552.92 2.92 456.8 93.2
66.9 600 597.82 2.18 504.22 95.78
70.4 650 659.03 9.03 572.21 77.79
72.9 700 705.14 5.14 626.26 73.74
75.5 750 755.59 5.59 688.49 61.51
77.8 800 802.69 2.69 749.82 50.18
79.9 850 848.07 1.93 812.22 37.78
82.5 900 908.04 8.04 900.31 0.31
84.1 950 947.45 2.55 962.14 12.14
86.0 1000 997.27 2.73 1045.49 45.49

| Page 11 of 15 |
Mathematical Exploration: Modelling the Heating Curve

87.9 1050 1051.16 1.16 1143.1 93.1

89.4 1100 1097.33 2.67 1233.99 133.99
91.0 1150 1151.18 1.18 1350.12 200.12
92.4 1200 1203.46 3.46 1475.62 275.62
93.9 1250 1267.11 17.11 1650.13 400.13
94.8 1300 1310.82 10.82 1787.93 487.93
95.7 1350 1360.84 10.84 1969.91 619.91
96.4 1400 1406.45 6.45 2166.96 766.96
97.0 1450 1453.39 3.39 2415.71 965.71
97.3 1500 1481.41 18.59 2597.69 1097.69
97.8 1550 1542.01 7.99 3158.16 1608.16
98.1 1600 1603.68 3.68 N/A N/A
Mean Difference 4.86 246.56

Legend

–•– Actual Data

4 1.1)!
––– 𝑓 "# (𝑥) = −<"1.1111!22 + 3544761.4331 + J1.1111335K

"#$%.!
$%
––– 𝑔"# (𝑥) = "1.11#23
#'(.)

Figure 4 – actual data against modeled functions

| Page 12 of 15 |
Mathematical Exploration: Modelling the Heating Curve

Conclusion
As the mean difference and the graph clearly indicates, the quadratic model 𝑓 "# (𝑥) is

clearly a more accurate model. Both of these models allow for the subtraction of constant c

from 𝑥 making both models useful in calculating the time required for milk to reach boiling

point for any given initial temperature.

In order to accurately estimate the time it would take for milk to reach a boil in the

particular stove, I would calculate the value of c based on the room temperature and

substitute the boiling point of milk which is 98.1 in the more accurate function 𝑓 "# (𝑥).

For example, taking the current room temperature of my kitchen which is 30.0 °C, I can

calculate the value of 𝑐 = 30.0 − 22.7 = 7.3. Then, Input the value of the boiling point of

milk which is 98.1 into function 𝑓 "# (𝑥) .

(𝑥 − 7.3) 0.092
𝑓 "# (𝑥) = −> + 3544761.4331 + 3 6
−0.0000277 0.0000554

(98.1 − 7.3) 0.092

𝑓 "# (98.1) = −> + 3544761.4331 + 3 6
−0.0000277 0.0000554

90.8 0.092
𝑓 "# (98.1) = −> + 3544761.4331 + 3 6
−0.0000277 0.0000554

90.8 0.092
−> + 3544761.4331 + 3 6 = 1144
−0.0000277 0.0000554

To the nearest second, it would take 1144 seconds for my milk to reach a boil when the
room temperature is 30.0 °C in the place where the stove is kept.

| Page 13 of 15 |
Mathematical Exploration: Modelling the Heating Curve

Limitations and Further Scope

Limitations
I believe one of the main limitations of my exploration was the limitations of my
experiment. For example, temperature fluctional in the room I conducted my experiment in
could have contributed to inaccurate data and hence resulting in an inaccurate
mathematical model. In order to reduce inaccuracies induced by variations in room
temperature, the experiment can be conducted in a more insulted environment for example
the recording room in our school which is has foam walls providing good thermal insulation
and minimizing room temperature variations. One other limitation of my experiment was
the lack of trials conducted, conducting more trials would have reduced random uncertainty
in the experiment data.

I think the limitation of the model I created was its restriction to be only useful on a certain
type of stove heating at a certain rate. This is because the rate at which the stove is heating
the liquid would change the heating curve making the function useless for accurately
predicting the time on a different stove. I think my model can be greatly improved is by
modifying the function in such way that the intensity of the heat source can be accounted
for in calculating time. Modifying the function in a such a way would make it much more
useful as it would be possible for one to modify the function based on the type of stove they
are using increasing the scope of its application.

Further Application
Other than determining the time it takes milk to reach a specific temperature on a stove,
the same method can be applied to different liquids in order to model their heating curve.
This can be useful for industrial applications as factories can easily determine the time a
certain heating process will take. The same can also be applied in restaurants where
automatic temperature tracking instruments are usually not available.

| Page 14 of 15 |
Mathematical Exploration: Modelling the Heating Curve

Bibliography
Desmos. 20 Jul 2018. Functions. https://support.desmos.com/hc/en-us/articles/207316093-
Functions. Accessed 21 Jul 2018.

Patricia Goodman. 2 Dec 2014. Add a trendline to a chart in Google Sheets.

https://service.goshen.edu/support/index.php?/default_import/Knowledgebase/Article/Vie
w/974/0/add-a-trendline-to-a-chart-in-google-sheets. Accessed 10 Jul 2018. Accessed 5 July
18.

Phil Clark. 12 Jan 2016. Differential Equations Newton's Law of Heating Problem.
https://www.youtube.com/watch?v=Kqy1xEUnEIE. Accessed 10 Jul 2018.

Sandra Lach Arlinghaus. 1994. CRC Press – Practical Handbook of Curve Fitting.

The University of British Columbia. Other differential equations. 7 May 1999

http://www.ugrad.math.ubc.ca/coursedoc/math100/notes/diffeqs/cool.html. Accessed 10
Jul 2018.

Vernier Software & Technology. Logger Pro 3 Quick Reference Guide. 9 May 2008
http://www2.vernier.com/manuals/LP3QuickRefManual.pdf. Accessed 4 July 2018.

| Page 15 of 15 |