Exploration Title: Minimising the packaging needed for

different 3D-shaped containers using optimisation.

Number of Pages: 20
Table of Contents

Introduction and Rationale.............................................................................................................................. 3

Aim and Methodology..................................................................................................................................3
Main Body..........................................................................................................................................................4
Optimising Cylindrical Products.................................................................................................................. 5
Cylindrical Container: Coke Can..................................................................................................................7
Optimising Conical Products........................................................................................................................9
Conical Container: Paper Cone.................................................................................................................. 12
Optimising Rectangular Prismatic Products...............................................................................................13
Rectangular Prismatic Container: Tetra Pak...............................................................................................17
Results Summary........................................................................................................................................ 18
Sensitivity Analysis.................................................................................................................................... 18
Cylindrical Containers......................................................................................................................... 18
Conical Containers............................................................................................................................... 19
Rectangular Prismatic Containers........................................................................................................ 19
Sensitivity Analysis Tabular Results and Analysis.............................................................................. 20
Conclusion........................................................................................................................................................21
Strengths and Limitations........................................................................................................................... 21
Further Research.........................................................................................................................................22
Bibliography.................................................................................................................................................... 23

2
​ ​ ​ Introduction and Rationale
In all markets selling tangible goods, packaging is vital for protecting the product, providing key

information, and promoting the business the product belongs to (Geddes, 2021). In 2024, there has

been a constant rise in packaging costs (KORPACK, 2024), meaning optimising the packaging

design on goods can significantly reduce a business’ long-term costs while allowing goods to be

properly packaged. Furthermore, this aligns with a business's sustainability and efficiency goals,

which are critical goals in today’s societies to implement to reduce the impact of climate change or

global warming.

As a Math and Business Management student, I have been taught the importance of sustainability,

and packaging being one of the best methods to implement it. The global packaging industry is a

massive contributor to environmental waste, with millions of tons generated yearly. This

exploration provides the opportunity not only to benefit the environment but also to align my

academic pursuits with my passion for sustainable practices.

Aim and Methodology

This exploration aims to investigate the minimal surface area of various containers and identify how

minimised surface area varies with tiny changes in the optimal dimensions. I will do this by, firstly,

minimising the surface area using a constant volume for all shapes to generate a derived formula.

Next, I applied that derived formula to the actual surface area of a real-life product that matches the

shape I optimised to calculate the wasted surface area percentage and use these results through

real-world data to make some sort of real-world approximation. I will conduct a sensitivity analysis

for all container shapes to identify changes in area with variable volume. Volume is constant to

container content preservation. Therefore, constrained optimisation concepts were used throughout

the exploration. In making any calculation, I utilised my graphical display calculator, and I

employed Desmos for graphing purposes.

3
The following symbols were used throughout the exploration and measured in cm.

Symbols Definition

𝑟 Radius

𝑟𝑜𝑝𝑡 Optimal dimensions radius

ℎ Height

ℎ𝑜𝑝𝑡 Optimal dimensions height

𝐴 Total surface area

𝐴𝑢𝑡𝑖𝑙𝑖𝑠𝑒𝑑 surface area used by a particular example

𝐴𝑜𝑝𝑡 Surface area after applying optimal dimensions

𝐴𝑤𝑎𝑠𝑡𝑒𝑑% Percent of wasted (additional surplus) surface

area when comparing the smallest possible area

with the surface area used by a specific example.

𝑉 Volume

𝑠 Slant height

𝑙 Length

𝑤 Width

Table 1. Table consisting of all symbols used in the exploration with definition

​ ​ ​ ​ Main Body

Due to a space limit for this project, only 1 example of real-life containers for each shape is shown,

affecting the results’ reliability, as other factors determine the 𝐴𝑤𝑎𝑠𝑡𝑒𝑑%. This investigation

optimises cylindrical products first, then conical, and then rectangular prismatic because I have

arranged these containers in the order of the top 3 most commonly shaped containers and from my

experience and from the least to most intricate in terms of math concepts. This allows exploration to
4
be relevant to the maximum number of containers while considering the limited space available.

Additionally, For this investigation’s purposes, we assume all 3D shapes’ real-world examples are

perfect representatives of the 3D shapes they are examples of.

Optimising Cylindrical Products

Figure 1. Dimensions of a cylinder

Firstly, we must use a cylinder's volume and surface area formulae to derive a formula that

minimises the surface area for cylinders. For cylinders radius and height are independent variables

as they are used in determining the dependent variable (minimised surface area).

2 +
𝑉 = π𝑟 ℎ; 𝑟, ℎ, 𝑉 ∈ ℝ (𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 1. 1)

2 +
𝐴 = 2π𝑟 + 2π𝑟ℎ; 𝐴 ∈ ℝ (𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 1. 2)

We isolate ℎ in equation 1.1 to allow only 1 of the 2 independent variables in terms of A, allowing

single-variable differentiation and the volume is constant, preventing it from being considered a

variable.

𝑉
​ ​ ​ ​ ℎ= 2 (𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 1. 3)
π𝑟

5
Substituting equation 1.3 into equation 1.2 allows a formula to be created for 𝐴 in terms of r and V.

​ ​ ​ ​ 𝐴 = 2π𝑟 + 2π𝑟
2
( ) 𝑉
π𝑟
2

Cancelling the 2π in the RHS (Right Hand Side):

2 2𝑉
​ ​ ​ ​ 𝐴 = 2π𝑟 + 𝑟

Now we differentiate for A in terms of r to find the stationary point, which allows identifying the

minimum surface area with constant volume. We differentiate by the power rule ( 𝑑
𝑑𝑟
𝑛 𝑛−1
𝑟 = 𝑛𝑟 )
where 𝑛 ∈ ℝ.

𝑑𝐴 2𝑉
​ ​ ​ ​ 𝑑𝑟
= 4π𝑟 − 2
𝑟

𝑑𝐴
𝑑𝑟
= 0 because stationary points have no gradient.

2𝑉
​ ​ ​ ​ 4π𝑟 − 2 = 0
𝑟

Finding a formula for r in terms of V to get critical or stationary points:

3
​ ​ ​ ​ 4π𝑟 − 2𝑉 = 0

𝑉 3
​ ​ ​ ​ 2π
=𝑟

3 3 𝑉
​ ​ 𝑉 = 2π𝑟 𝑜𝑝𝑡
(𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 1. 4) 𝑟𝑜𝑝𝑡 = 2π
(𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 1. 5)

I need to ensure that this stationary point is a minimum because the radius and height will be

aligned to create the optimal dimensions for the smallest surface area with constant volume,

connecting with the exploration’s objective. This can be proven by differentiating again:

2
𝑑𝐴 4𝑉
​ ​ ​ ​ ​ ​ 2 = 4π + 3
𝑑𝑟 𝑟 𝑜𝑝𝑡

6
Use equation 1.4 to substitute with V to identify the stationary point’s nature (verify if it’s a

minimum):

3
2
𝑑𝐴 4(2π𝑟 )
​ ​ ​ ​ 2 = 4π + 3
𝑜𝑝𝑡

𝑑𝑟 𝑟 𝑜𝑝𝑡

3
Cancel the 𝑟 𝑜𝑝𝑡
:

2
𝑑𝐴
​ ​ ​ ​ 2 = 4π + 8π
𝑑𝑟

2
𝑑𝐴
​ ​ ​ ​ 2 = 12π
𝑑𝑟

​ ​ ​ ​ 12π > 0 ∴ it proves that this point is a minimum because positive

concavity corresponds to a minimum while negative concavity corresponds to a maximum.

To identify a relationship between the dimensions of the cylinder, we equate equations 1.1 and 1.4.

3 2
​ ​ ​ ​ ​ 2π𝑟 = π𝑟 ℎ

ℎ𝑜𝑝𝑡 = 2𝑟𝑜𝑝𝑡 (𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 1. 6)​

The result of the height being double the radius identifies the cylinder’s optimal dimensions. This

result was expected as, according to equations 1.1 and 1.2, the radius affects a cylinder’s surface

area more greatly than height, if height was double the radius, then the surface area would be

minimal with constant volume.

7
 Cylindrical Container: Coke Can

Figure 2. Dimensions of a coke can

Our calculations can be applied to this example to identify how close Coca-Cola’s bottle is to matching the

smallest possible surface area with its current volume. Firstly, use equation 1.5 to find the optimal radius:

3 393
​ ​ ​ ​ 𝑟𝑜𝑝𝑡 = 2π
≈ 3. 97𝑐𝑚 (𝑟𝑜𝑢𝑛𝑑𝑒𝑑 𝑡𝑜 2 𝑑𝑒𝑐𝑖𝑚𝑎𝑙 𝑝𝑙𝑎𝑐𝑒𝑠)

Minimum surface area with the optimal radius for the Coke Can can be confirmed visually by Desmos:

​ ​ Figure 3. Graphing cylinder surface area equation and its derivative for visualisation, graphed from Desmos.

The radius was put as “x” in Desmos for easier graphing. This graph contains both the equation 1.2 and its

derivative. In the x-coordinate, the graph shows the optimal radius, and the y-coordinate of 𝑓(𝑥) shows the

minimum surface area with constant volume.

Use equation 1.6 to find the optimal height and confirm this with equation 1.3:

​ ​ ​ ​ ℎ𝑜𝑝𝑡 = 2(3. 97) = 7. 94𝑐𝑚

393
​ ​ ​ ​ ℎ𝑜𝑝𝑡 = 2 ≈ 7. 9419 = 7. 94𝑐𝑚
π(3.97)

8
The difference of 0.01cm in optimised height is negligible because 0.01 makes a tiny impact on the surface

area minimisation for any container, this also applies to conical and rectangular prismatic containers.

Use these values and equation 1.2 to find the minimum surface area and the wasted surface area percent:

2 2
​ ​ ​ ​ 𝐴𝑜𝑝𝑡 = 2π(3. 97) + 2π(3. 97)(7. 94) ≈ 297. 09𝑐𝑚

𝐴𝑢𝑡𝑖𝑙𝑖𝑠𝑒𝑑−𝐴𝑜𝑝𝑡𝑖𝑚𝑖𝑠𝑒𝑑
​ ​ ​ ​ 𝐴𝑤𝑎𝑠𝑡𝑒𝑑% = 𝐴𝑢𝑡𝑖𝑙𝑖𝑠𝑒𝑑
· 100

For the fixed volume, the additional surplus area caused between original area of Coke can vs the theoretical

minimum area is the percent of wasted area. It is calculated as a percentage because of the unequal

dimensions of each real-life container for each shape. This means that the surplus area is more likely to be

higher for containers with larger dimensions, creating assumptions to identify which container has the largest

wasted area without considering its size, leading to inaccurate results.

306.87−297.09
​ ​ ​ ​ 𝐴𝑤𝑎𝑠𝑡𝑒𝑑% = 306.87
· 100 ≈ 3. 19%

𝐴𝑤𝑎𝑠𝑡𝑒𝑑% signifies a small percentage of unnecessary area within the bottle. This results in significant

material costs caused by a less efficient surface area to volume ratio, due to the additional surplus area added

to the can to provide the same volume if that surplus area wasn’t added to the can. If Coke cans didn’t have

this additional surplus area, the material costs would’ve been cheaper while having constant volume. Since

an estimated 1.9 billion Coke cans (The Coca-Cola Company, n.d.) are sold daily, this change adds to an

enormous wasted area. More specifically, the total wasted area would be:

9 10 2
​ ​ ​ ​ 1. 9 · 10 · (306. 87 − 297. 09) = 1. 86 · 10 𝑐𝑚

This massively harms sustainability because sustainability’s main premise is to minimise material wastage.

9
​ ​ ​ Optimising Conical Products

Figure 4. Dimensions of a cone

We do the same process as finding the derived formula for minimising surface area for a cylinder. Since this

is a similar process compared to the cylindrical one, not every step will be shown.

2 2 2 +
𝐴 = π𝑟 + π𝑟𝑠, 𝑠 = 𝑟 + ℎ ; 𝑟, ℎ, 𝐴 ∈ ℝ ​ 𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 2. 1
2
π𝑟 ℎ +
𝑉= 3
; 𝑉 ∈ℝ 𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 2. 2

Isolating h in equation 2.2 to allow 1-variable differentiation:

3𝑉
​ ​ ​ ​ ​ ℎ= 2 ​ ​ ​ ​ 𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 2. 3
π𝑟

Substituting equation 2.3 into equation 2.1 and simplifying to generate a formula for r in terms of A:

2
​ ​ ​ ​ 𝐴 = π𝑟 + π𝑟
2 2
𝑟 + ( ) 3𝑉
π𝑟
2

​ ​ ​ ​ 𝐴 = π𝑟 + π𝑟
2
( 2 6
π 𝑟 +9𝑉

π𝑟
2 4
2

)
2 6 2
2 π 𝑟 +9𝑉
​ ​ ​ ​ 𝐴 = π𝑟 + 𝑟

Now differentiate the derived area formula in terms of the radius to find the stationary point/s. To find the

( ( )−𝑢( )
)
𝑑𝑢 𝑑𝑣
𝑑𝐴 𝑣
derivative, the quotient rule is required for the 2nd term 𝑑𝑟
= 𝑑𝑟
2
𝑑𝑟

1
𝑑𝐴 𝑢 2 6 2 2 6 2
​ ​ ​ 𝑑𝑟
= 2π𝑟 + 𝑣
𝑤ℎ𝑒𝑟𝑒 𝑢 = π 𝑟 + 9𝑉 𝑜𝑟 (π 𝑟 + 9𝑉 ) 2 𝑎𝑛𝑑 𝑣 = 𝑟

10
Now, we use chain rule, ( 𝑑𝐴
𝑑𝑟
=
𝑑𝐴
𝑑𝑤
·
𝑑𝑤
𝑑𝑟 ) where 𝑤 = π 𝑟 2 6 2
+ 9𝑉 in this context. Finding function for
𝑑𝐴
𝑑𝑟
:

1 2 5
𝑑𝑢 1 2 5 2 6 2 −2 𝑑𝑢 3π 𝑟 𝑑𝑣
​ ​ ​ 𝑑𝑟
= 2
· 6π 𝑟 · (π 𝑟 + 9𝑉 ) ∴ 𝑑𝑟
= 2 6 2
and 𝑑𝑟
=1
π 𝑟 +9𝑉

​ ​ ​
𝑑𝐴
= 2π𝑟 +
( 3π 𝑟
2 6
2 6

π 𝑟 +9𝑉
2 )(
− 1( π 𝑟 +9𝑉 )
2
2 6 2
)
𝑑𝑟 𝑟

2 6 2
To simplify, in 2nd term, we combine fractions in numerator and multiply denominator with π 𝑟 + 9𝑉 :
2 6 2
𝑑𝐴 2π 𝑟 −9𝑉
​ ​ ​ 𝑑𝑟
= 2π𝑟 + 2 2 6 2
𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 2. 4
𝑟 π 𝑟 +9𝑉

𝑑𝐴
After equating 𝑑𝑟
= 0, remove the denominator by multiplying it by 0.

3 2 6 2 2 6 2
​ ​ ​ 2π𝑟 π 𝑟 + 9𝑉 + 2π 𝑟 − 9𝑉 = 0

3 2 6 2 2 6 2
​ ​ ​ − 2π𝑟 π 𝑟 + 9𝑉 = 2π 𝑟 − 9𝑉

Square LHS and RHS and simplify to generate a formula for r in terms of V for a stationary point:

4 12 2 6 2 4 12 2 6 2 4
​ ​ ​ 4π 𝑟 + 36π 𝑟 𝑉 = 4π 𝑟 − 36π 𝑟 𝑉 + 81𝑉

2 6 2 3
72π 𝑟 𝑉 2π𝑟 2
​ ​ ​ ​ 𝑉= 2 ∴𝑉 = 3
𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 2. 5
81𝑉

Verifying that equation 2.5 represents a minimum point by differentiating equation 2.4:

2 6 2 𝑑𝑢 2 5
𝑢 = 2π 𝑟 − 9𝑉 ∴ 𝑑𝑟
= 12π 𝑟

1
2 2 6 2
( )
2
𝑣 = 𝑟 π 𝑟 + 9𝑉

The product rule (𝑥𝑤' + 𝑤𝑥') is required along with chain rule to differentiate 𝑣
1
2 5
2 2 6 2
𝑑𝑤
( ) 𝑑𝑥 3π 𝑟
2
​ ​ ​ 𝑤 =𝑟 , 𝑑𝑟
= 2𝑟 ; 𝑎𝑛𝑑 𝑥 = π 𝑟 + 9𝑉 , 𝑑𝑟
= 2 6 2
π 𝑟 +9𝑉

1
2 7
2 6 2
𝑑𝑣
( ) 3π 𝑟 𝑑𝑥
2
​ ​ ​ 𝑑𝑟
= 2𝑟 π 𝑟 + 9𝑉 + 2 6 2
; Already multiplied 𝑤 to 𝑑𝑟
π 𝑟 +9𝑉

Applying the quotient rule and simplifying:

​ ​ ​
𝑑𝐴
2

2 = 2π +
12π 𝑟 𝑟
2 5
( 2) 2 6 2
π 𝑟 +9𝑉 − 2π 𝑟 −9𝑉 ( 2 6 2
) 2𝑟( 2
2 6
π 𝑟 +9𝑉 +
2
2 7
3π 𝑟
2 6
π 𝑟 +9𝑉
2 )​
(( )
1
𝑑𝑟 2 2 6 2
)
2
𝑟 π 𝑟 +9𝑉

11
​ ​ ​
𝑑𝐴
2

2 = 2π +
12π 𝑟
2 7
π 𝑟 +9𝑉 − (
2 6 2
2 6 (2
2 6
2π 𝑟 −9𝑉 )(5π 𝑟 +18𝑟𝑉 )
2 2 6

π 𝑟 +9𝑉
2

)
𝑑𝑟 2 10 4 2
π 𝑟 +9𝑟 𝑉

4 13 2 7 2 4
2π 𝑟 +117π 𝑟 𝑉 +162𝑟𝑉
2
𝑑𝐴 2 6 2

​ ​ ​ 2 = 2π + π 𝑟 +9𝑉
2 10 4 2
𝑑𝑟 π 𝑟 +9𝑟 𝑉

2 4 12 2 6 2 4
= 2π + ( )
𝑑𝐴 𝑟 2π 𝑟 +117π 𝑟 𝑉 +162𝑉
​ ​ ​ 2
𝑑𝑟 4 2 6 2 2 6 2
(
𝑟 π 𝑟 +9𝑉 ) π 𝑟 +9𝑉

3
3 2 6 2 4 12 2 6 2 4
( )
2
2π𝑟 π 𝑟 +9𝑉 +2π 𝑟 +117π 𝑟 𝑉 +162𝑉
​ ​ ​ 3 > 0 ∴ This is a minimum point
3 2 6 2
( )
2
𝑟 π 𝑟 +9𝑉

The final expression is larger than 0 because 𝑟 and 𝑉 were already declared earlier as positive real numbers

and all terms and constants are being added to each other.​

Substitute equation 2.5 into equation 2.3 and isolate h to generate the optimal dimensions that minimise area:

( )
3
2π𝑟 2
3 3
​ ​ ​ ​ ℎ= 2
π𝑟

3
2π𝑟 2
​ ​ ​ ​ ℎ= 2
π𝑟

3 3𝑉
​ ​ ℎ𝑜𝑝𝑡 = 2 2 · 𝑟𝑜𝑝𝑡 𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 2. 6, 𝑟𝑜𝑝𝑡 = 𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 2. 7
2π 2

This result was unexpected as I believed the optimal dimensions relation for cylinder would have a larger

height to radius ratio than the cone. This was due to my thought of radius impacting cylinder more

significantly than cone compared to height, as there was a larger presence of height in cone.

12
​ ​ ​ ​ Conical Container: Paper Cone

Figure 5. COMATEC. (2023). PAPER CONE [Photograph]. Comatec.

https://www.comatec.fr/gb/1569-paper-cone.html

We incorporate a similar procedure to identify the wasted area of this paper cone as my water bottle.

Substituting V into equation 2.7 to find the optimal radius and the height by optimal radius substitution:

3 3(89.80)
​ ​ ​ ​ 𝑟𝑜𝑝𝑡𝑖𝑚𝑖𝑠𝑒𝑑 = ≈ 3. 12𝑐𝑚
2π 2

​ ​ ​ ​ ℎ𝑜𝑝𝑡𝑖𝑚𝑖𝑠𝑒𝑑 = 2 2 · 3. 12 ≈ 8. 82𝑐𝑚

Using Desmos to graph the cone’s total area function in terms of radius with its derivative:

Figure 6. Graphing cone surface area equation and its derivative for visualisation, graphed from Desmos.

Utilising equation 2.2 to find 𝐴𝑜𝑝𝑡𝑖𝑚𝑖𝑠𝑒𝑑 and using that to find 𝐴𝑤𝑎𝑠𝑡𝑒𝑑 :

2 2 2 2
​ ​ ​ ​ 𝐴𝑢𝑡𝑖𝑙𝑖𝑠𝑒𝑑 = π · (3. 12) + 3. 12π 3. 12 + 8. 82 ≈ 122. 28𝑐𝑚

13
 124.54−122.28
​ ​ ​ ​ 𝐴𝑤𝑎𝑠𝑡𝑒𝑑% = 124.54
· 100 ≈ 1. 81%

This is significantly lower 𝐴𝑤𝑎𝑠𝑡𝑒𝑑% compared to the cylindrical container, meaning the paper cone

is already quite close to its minimum surface area. Paper cone can be considered as a

mass-produced product because in clubhouses, they’re filled with water, drank from, and thrown

away. The low additional surplus area in paper cones reduces sustainability by a lower amount,

especially when the paper cone market is valued at $8. 05 million in 2023 (Coherent Market

Insights, 2023) and expected to reach $13. 18 million in 2030. Finding projected percent growth:

13.18−8.05
​ ​ ​ ​ ​ 13.18
· 100 ≈ 38. 92%

This is a very significant percent growth, reinforcing the impact on sustainability from paper cones.

Optimising Rectangular Prismatic Products

Figure 7. Dimensions of a rectangular prism

In deriving formulae for cylindrical and conical containers, there were only 2 unknowns, causing me to write

their area in terms of 1 unknown and optimise by 1-variable differentiation. However, in rectangular prisms,

there are 3 unknowns, meaning optimisation by 1-variable differentiation won’t work. This means

multivariable calculus is needed. More specifically, Lagrange Multipliers, as it’s a technique that optimises a

function with multiple unknowns when there is a constraint using partial derivatives. The constraints look

similar to this (Khan Academy, n.d.):

𝑔(𝑥, 𝑦,...) = 𝑘; where 𝑘 is some constant and 𝑔 is a multivariable function.

14
In this context, 𝑔 has the same input space as 𝑓 where 𝑓 and 𝑔 are both multivariable functions. The goal is

to find the points in which the contour lines are tangential, which is equivalent to finding points where the

gradient vectors 𝑓 and 𝑔 are parallel to each other (Khan Academy, n.d.).

+
𝑉 = 𝑤ℎ𝑙; 𝑤, ℎ, 𝑙, 𝑉 ∈ ℝ 𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 3. 1

+
𝐴 = 2𝑙𝑤 + 2ℎ𝑙 + 2ℎ𝑤 = 2(𝑙𝑤 + ℎ𝑙 + ℎ𝑤); 𝐴 ∈ ℝ 𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 3. 2

𝑓(𝑤, ℎ, 𝑙) = 2𝑙𝑤 + 2ℎ𝑙 + 2ℎ𝑤; The multivariable function 𝑓 represents the unknowns involved

​ 𝑔(𝑤, ℎ, 𝑙) = 𝑉; The multivariable function 𝑔 represents the constraint

∇𝑓 = λ∇𝑔

∇ is a multivariable function’s gradient, meaning it represents all its partial derivatives. ∇𝑓 contains:

​ ​ ​ ​ ∇𝑓 = ( ∂𝐴
∂𝑤
,
∂𝐴
∂ℎ
,
∂𝐴
∂𝑙 )
Partial derivatives are simply when we differentiate a function in terms of specific variables, meaning the

variable/s not part of the “specific variables” are considered constants, meaning that they become 0 after

partial differentiation if there are no “specific variables” with that other variable.

Performing partial derivatives:

∂𝐴
∂𝑤
= 2𝑙 + 2ℎ = λ(ℎ𝑙) (1)

∂𝐴
∂ℎ
= 2𝑙 + 2𝑤 = λ(𝑤𝑙) (2)

∂𝐴
∂𝑙
= 2ℎ + 2𝑤 = λ(ℎ𝑤) (3)

λ is the Lagrange Multiplier, a constant that equalises the RHS with the partial derivative of the area

multivariable function. Now these become simultaneous equations, solving these equations:

Multiply (1) with 𝑤, multiply (2) with ℎ, and multiply (3) with 𝑙.

(2𝑙 + 2ℎ) · 𝑤 = λ(𝑤ℎ𝑙)

(2𝑙 + 2𝑤) · ℎ = λ(𝑤ℎ𝑙)

(2ℎ + 2𝑤) · 𝑙 = λ(𝑤ℎ𝑙)

We can equate all equations now as RHS is the same.

(2𝑙 + 2ℎ) · 𝑤 = (2𝑙 + 2𝑤) · ℎ = (2ℎ + 2𝑤) · 𝑙

For simplicity, equate (1) and (2) first and simplify.

2𝑙𝑤 + 2ℎ𝑤 = 2ℎ𝑙 + 2ℎ𝑤

15
 2𝑤 = 2ℎ ∴ 𝑤 = ℎ

Now equate (2) and (3) and simplify.

2𝑤𝑙 + 2ℎ𝑤 = 2ℎ𝑙 + 2𝑤𝑙

2𝑤 = 2𝑙 ∴ 𝑤=𝑙 ∴

3
𝑤=ℎ=𝑙= 𝑉 𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 3. 3

To prove that this result is a minimum, we can’t differentiate again because there are 3 unknowns, meaning

the Jacobian and Hessian matrices will be needed, which specialise in proving the point’s status. Firstly, we

isolate w in equation 3.1:

𝑉
​ ​ ​ ​ ​ ​ 𝑤= ℎ𝑙
​

Substitute this into equation 3.2:

​ ​ ​ ​ ​ 𝐴=2 𝑙 ( ( ) + ℎ𝑙 + ℎ( ))
𝑉
ℎ𝑙
𝑉
ℎ𝑙

​ ​ ​ ​ ​ 𝐴=2 ( 𝑉
ℎ
+ ℎ𝑙 +
𝑉
𝑙 )
2𝑉𝑙+2𝑉ℎ 2𝑉(ℎ+𝑙)
​ ​ ​ ​ ​ 𝐴= ℎ𝑙
+ 2ℎ𝑙 ∴ 𝐴 = ℎ𝑙
+ 2ℎ𝑙 𝐸𝑞𝑢𝑎𝑡𝑖𝑜𝑛 3. 4

Now we implement the Jacobian Matrix to find the partial derivatives of equation 3.4:

∂𝐴 ∂𝐴
​ ​ ​ ​ ​ ​ 𝐽 = ⎡ ∂ℎ , ⎤
⎣ ∂𝑙 ⎦

After this, we apply the Hessian Matrix to identify the minimum using the Jacobian Matrix.
2 2
∂𝐴
⎡ ∂ℎ
2
∂𝐴
∂ℎ∂𝑙
⎤
​ ​ ​ ​ ​ 𝐻 =⎢ ⎥
⎢ 2
∂𝐴 ∂𝐴
2
⎥
⎣ ∂𝑙∂ℎ ∂𝑙
⎦

Since this above matrix is symmetric, according to Sylvester’s Criterion “A symmetric matrix A is positive

definite if and only if all principal minors are positive. If A is positive definite all special minors are

positive.” (math.washington, n.d.). In simple terms, positive definite are conditions that allow for easier

calculations, principle minors are smaller matrix determinants that arise from the top left of matrices while

special minors are all other matrix determinants from a larger matrix.

1st and 2nd Derivatives of the Jacobian and Hessian Matrices respectively

16
𝐴 = 2ℎ𝑙 +
2𝑉ℎ+2𝑉𝑙 From equation 3.5, differentiate each term:
ℎ𝑙
∂𝐴 2𝑉 ∂𝐴 2𝑉
2𝑉 2𝑉 ∂𝑙
= 2ℎ − 2 + 0∴ ∂𝑙
= 2ℎ − 2
𝐴 = 2ℎ𝑙 + 𝑙
+ ℎ
𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 3. 5 𝑙 𝑙

∂𝐴 2𝑉 ∂𝐴 2𝑉
∂ℎ
= 2𝑙 + 0 − 2 ∴ ∂ℎ
= 2𝑙 − 2
ℎ ℎ

2 2 2 2
∂𝐴 4𝑉 ∂𝐴 4𝑉 ∂𝐴 4𝑉 ∂𝐴 4𝑉
2 =0+ 3 ∴ 2 = 3 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 3. 6 2 =0+ 3 ∴ 2 = 3 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 3. 7
∂ℎ ℎ ∂ℎ ℎ ∂𝑙 𝑙 ∂𝑙 𝑙

2 2 2 2
∂𝐴 ∂𝐴 ∂𝐴 ∂𝐴
∂ℎ∂𝑙
= 2 + 0∴ ∂ℎ∂𝑙
=2 ∂𝑙∂ℎ
= 2 + 0∴ ∂𝑙∂ℎ
=2

Table 2. Table consisting of 1st and 2nd derivatives of Jacobian and Hessian Matrices.

4𝑉
⎡ 3 2 ⎤
​ ​ ​ ​ ​ 𝐻 =⎢ ℎ
4𝑉 ⎥
⎢ 2 ⎥
⎣ 𝑙
3
⎦
3
According to equation 3.3, ℎ = 𝑙 = 𝑉 we can substitute equation 3.3 into equations 3.6 and 3.7:

4𝑉 4𝑉
​ ​ ​ ​ ​ 3 ∴ 𝑉
= 4
( 𝑉)
3

4 2
​ ​ ​ ​ ​ 𝐻 = ⎡⎢ ⎤⎥
⎣ 2 4 ⎦

Now, we find the determinant of the matrix to identify the point’s status:

​ ​ ​ ​ ​ 𝐻𝑑𝑒𝑡 = 4(4) − 2(2) = 12

​ ​ ​ ​ ​ 12 > 0 and 12 > 4

According to Sylvestor’s criterion, this point is a minimum because the first comparison of the

determinant with 0 identifies whether the determinant is negative or positive. The 2nd comparison

compares the determinant with the 1st element in the Hessian Matrix. This point would've been a

maximum if it was smaller than the 1st element.

17
 Rectangular Prismatic Container: Tetra Pak

Figure 8. MARINELAND. (n.d.). Standard Black Retangular Aquarium [Photograph]. marineland.

https://www.marineland.com/products/aquariums/standard-black-retangular-aquarium.aspx

Firstly, we use equation 3.3 to calculate the optimal dimensions of the aquarium.

3
​ ​ ​ ​ ​ 3
𝑉𝑢𝑡𝑖𝑙𝑖𝑠𝑒𝑑 ∴ 379. 5 ≈ 7. 24𝑐𝑚

Now, we use these optimal dimensions to identify the optimised area using equation 3.2 and find the

wasted area by calculating the difference between utilised area and optimised area:

2 2 2 2
​ ​ ​ 𝐴𝑜𝑝𝑡𝑖𝑚𝑖𝑠𝑒𝑑 = 2(7. 24 + 7. 24 + 7. 24 ) ≈ 314. 51𝑐𝑚

Due to the 3D dimensions needed for graphing, the graph wasn’t possible for this container.

379.50−314.51
​ ​ ​ 𝐴𝑤𝑎𝑠𝑡𝑒𝑑% = 379.50
· 100 ≈ 17. 13%

This 𝐴𝑤𝑎𝑠𝑡𝑒𝑑% is high, indicating large surplus area per tetra pak container. Since it’s

mass-produced, this harms sustainability significantly more as 1 container has an additional 17.13%

of surplus area for constant volume, creating an inefficient surface area to volume ratio. Total tetra

pak containers sold are 190 million (Tetra Pak, n.d.), meaning the wasted area is around:

8 10 2
​ ​ ​ 1. 9 · 10 · (379. 5 − 314. 51) = 1. 23 · 10 𝑐𝑚

This result assumes that all 190 million tetra paks sold are the same size and that this container is a

perfect rectangular prism. Throughout the results of all 3 containers, I learned the importance of

optimisation in the real world, mainly with Lagrange Multipliers because they optimise when

multiple unknowns and constraints, hence, more applicable to real-world situations.

18
 Results Summary

Table 3. Optimal dimensions for all 3D shapes

Cylindrical Conical Rectangular Prismatic
Optimal dimensions ℎ𝑜𝑝𝑡 = 2𝑟𝑜𝑝𝑡 ℎ𝑜𝑝𝑡 = 2 2 𝑟𝑜𝑝𝑡 𝑤=ℎ=𝑙= 𝑉
3

Table 4. Wasted area percent for all 3D shape examples.

Coke Can Paper Cone Tetra Pak
Wasted area percent 3. 19% 1. 81% 17. 13%

​ ​ ​ ​ Sensitivity Analysis

Sensitivity analysis is a quantitative technique that explains how changes in the container dimensions affect

the surface area (Castillo et al, 2008). By assessing how dimensions affect area, we can identify how

packaging is affected by changes in volume for each container shape, identifying which container shape is

the most robust and efficient. The dimensions will be changed by 1%, 5%, 10%, and 20% to interpret the

result alterations with each % and simply interpret the physical relationship between dimensions and area.

​ ​ ​ ​ Cylindrical Containers

Firstly, substitute cylinder optimal dimensions in the total surface area formula for the cylinder and simplify:

2
​ ​ ​ ​ ​ 𝐴 = 2π𝑟 + 2π𝑟(2𝑟)

2 2 2
​ ​ ​ ​ ​ 𝐴 = 2π𝑟 + 4π𝑟 = 6π𝑟

Differentiate to get the result:

𝑑𝐴
​ ​ ​ ​ ​ 𝑑𝑟
= 12π𝑟 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 4. 1

Performing sensitivity analysis by substituting 𝑟𝑜𝑝𝑡 in equation 4.1:

𝑑𝐴
​ ​ ​ ​ ​ 𝑑𝑟
= 12π(5. 68) ≈ 214. 13

Increasing and decreasing 5.68 by 1% and substituting this value in equation 4.1:

𝑑𝐴
​ ​ ​ 5. 68 · 1. 01 ≈ 5. 74𝑐𝑚 ∴ 𝑑𝑟
= 12π(5. 74) ≈ 216. 27

𝑑𝐴
​ ​ ​ 5. 68 · 0. 99 ≈ 5. 62𝑐𝑚 ∴ 𝑑𝑟
= 12π(5. 62) ≈ 211. 99

Converting results to percentage:

19
 𝑓𝑖𝑛𝑎𝑙 𝑣𝑎𝑙𝑢𝑒−𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑣𝑎𝑙𝑢𝑒
​ ​ ​ 𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝐶ℎ𝑎𝑛𝑔𝑒 = 𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑣𝑎𝑙𝑢𝑒
· 100

216.27−214.13
​ ​ ​ 𝐼𝑛𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔 5. 68: 214.13
· 100 = 0. 9994%

Decreasing 5.68 will result in the negative of the same percentage as increasing 5.68 ∴ − 0. 9994%.

​ ​ ​ ​ Conical Containers

Substituting optimal dimensions into the total surface area formula for cones:

​ ​
2
𝐴 = π𝑟 + π𝑟
( 2
𝑟 + (2𝑟 2)
2

) = 𝐴 = π𝑟 + π𝑟
2
( 2 2
𝑟 + 8𝑟 )
​ ​ ​ ​ ​
2
𝐴 = π𝑟 + π𝑟 ( )
9𝑟
2
= 4π𝑟
2

Differentiate:

𝑑𝐴
​ ​ ​ ​ ​ 𝑑𝑟
= 8π𝑟 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 4. 2

Performing sensitivity analysis by using 𝑟𝑜𝑝𝑡𝑖𝑚𝑖𝑠𝑒𝑑 as the baseline value in equation 4.2:

𝑑𝐴
​ ​ ​ ​ ​ 𝑑𝑟
= 8π(3. 12) ≈ 78. 41

Increasing and decreasing 3.12 by 1% and substituting this value in equation 4.2:

𝑑𝐴
​ ​ ​ 3. 12 · 1. 01 ≈ 3. 15𝑐𝑚 ∴ 𝑑𝑟
= 8π(3. 15) ≈ 79. 20

𝑑𝐴
​ ​ ​ 3. 12 · 0. 99 ≈ 3. 09𝑐𝑚 ∴ 𝑑𝑟
= 8π(3. 09) ≈ 77. 63

Converting results into percentage:

79.20−78.41
𝐼𝑛𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔 3. 12: 79.20
· 100 = 1. 0075%

Decreasing 3.12 will result in the negative of the same percentage as increasing 3.12 ∴ − 1. 0075%.

​ ​ ​ Rectangular Prismatic Containers

3
Substituting 𝑉 into the total surface area formula for rectangular prisms:

2 2 2
𝐴 = 2(( 𝑉) + ( 𝑉) + ( 𝑉) )
3 3 3
​ ​ ​ ​

Convert the result into exponential form to ease differentiation:

2
2
6( 𝑉) = 6𝑉
3
​ ​ ​ ​ 3

20
Differentiate:​ ​ ​ ​ ​
1
𝑑𝐴 −3 4
𝑑𝑟
= 4𝑉 = 3 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 4. 3
𝑉

Performing sensitivity analysis by using optimal dimensions as the baseline value in equation 4.3:
𝑑𝐴 4
​ ​ ​ ​ ​ 𝑑𝑟
= 42.27
≈ 0. 0946 (𝑟𝑜𝑢𝑛𝑑 𝑡𝑜 4 𝑑. 𝑝.)

Increasing and decreasing 5.68 by 1% and substituting this value in equation 4.1:
𝑑𝐴 4
​ ​ ​ 42. 27 · 1. 01 ≈ 42. 69𝑐𝑚 ∴ 𝑑𝑟
= 42.69
≈ 0. 0937

𝑑𝐴 4
​ ​ ​ 42. 27 · 0. 99 ≈ 41. 84𝑐𝑚 ∴ 𝑑𝑟
= 41.84
≈ 0. 0956

Converting results into percentage:

0.0937−0.0946
𝐼𝑛𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔 42. 27: 0.0946
· 100 =− 0. 9514%

Decreasing 42.27 will result in the positive of the same percentage as increasing 42.27 ∴ 0. 9514%

To enhance concision, the same process for changing the optimal dimensions by 5%, 10%, and 20%

will be done without working out and the whole result will be shown as a table.

Sensitivity Analysis Tabular Results and Analysis

Cylindrical Conical Rectangular Prismatic

1% Increase 0.9994% 1.0075% -0.9514%
1% Decrease -0.9994% -1.0075% 0.9514%
5% Increase 5.0003% 5.0056% -4.7319%
5% Decrease -4.9994% -4.9950% 5.2963%
10% Increase 10.0004% 10.0058% -9.0623%
10% Decrease -9.9995% -9.9952% 11.1461%
20% Increase 20.0006% 20.0064% -16.6405%
20% Decrease -19.9996% -19.9958% 25.0393%
Table 5. Table displaying all sensitivity analysis results

This table represents the results of the sensitivity analysis when optimal dimensions were altered by

± 1%, ± 5%, ± 10%, and ± 20%. Rectangular prismatic containers represent the highest

sensitivity to changes in optimal dimensions for the 20% decrease, despite having the lowest

sensitivity for the other percents and conical having the highest sensitivity for the other percents.

This may be due to the absence of exponential terms in the total surface area formula for

21
rectangular prism, unlike cones and cylinders, causing non-linear behavior. This also explains the

higher sensitivity for cylinders than cones as cylinders have less exponential terms than cones.

Additionally, the reduction in surface area for rectangular prismatic containers if volume rises is

counterintuitive because normally changes in volume increase surface area due to the dimensions

being affected in both.

​ ​ ​ ​ ​ Conclusion
The results reveal Coke cans are one of the most sold products worldwide, making its 3.19%

surplus area significant. Even tiny decreases in this percent can immensely improve sustainability

and cost-effectiveness and the same point applies to the tetra pak containers but to a smaller extent

in the way that it isn’t as widely sold as Coke cans. Still, it has a significantly larger surplus area per

container which indicates massive optimisation potential for tetra pak to benefit sustainability.

Paper cones have the least surplus area with a growing market, indicating the highest sustainability

and closest to optimal design efficiency for paper cones. Furthermore, the sensitivity analysis

revealed the changing surface area rate with 1%, 5%, 10%, and 20% in optimal dimensions where

the 20% decrease observed an anomalous result where the rectangular prism showed an abnormally

large surface area change and same for cylinder but to a smaller extent, causing the cone to have the

lowest surface area change when for other sensitivities, it had the highest alterations.

​ ​ ​ ​ Strengths and Limitations

There are numerous strengths in this exploration. The most important one is its relevance to

real-world problems, as maximising sustainability and material efficiency has been an enormous

challenge worldwide. Another strength is a variety of approaches were applied to maximise the

results' reliability. Differentiation, sensitivity analysis, and identifying the percentage wasted area of

each real-life object with approximate calculations on total surplus area corresponding to a specific

22
3D shape were applied. Reliability is enhanced because integrating multiple concepts for surface

area minimisation provides more details on the results.

However, there are a few limitations to this exploration. Firstly, all containers were assumed to be

perfect examples of the investigated 3D shapes, but these examples can have irregularities, meaning

that the surplus area percent may not be completely accurate due to the potential alterations in

optimal surface area. Another limitation is that sensitivity analysis only remains accurate if optimal

dimensions don’t change. In real-world situations, this is false due to specific design requirements,

for instance, if a manufacturer can’t construct specific dimension/s for a container, then the entire

container’s dimensions must be changed to change that specific dimension/s, lowering the

reliability of sensitivity analysis. Another limitation is that I utilised various external sources for

various purposes. The issue is that the validity or reliability of these sources isn’t proven anywhere,

and some of them like the sales of Coke are estimates, creating uncertainty on the results.

​ ​ ​ ​ Further Research
A related extension to this investigation could be to explore irregular container shapes as they have

more practical applications. This allows more relevant results and directly correlates with

sustainability and the amount of packaging required when the area of an irregular-shaped container

is minimised with constant volume. AI methods should be used to calculate the optimal dimensions

of these containers due to the high intricacy of identifying the optimal dimensions of irregular

shapes. Lastly, more real-life containers need to be experimented on to improve the reliability of the

results regarding the wasted surface area, as 1 example is too little.

23
​ ​ ​ ​ ​ Bibliography
Castillo, E., Mínguez, R., & Castillo, C. (2008). Sensitivity analysis in

optimization and reliability problems. Reliability Engineering & System

Safety, 93(12), 1788-1800.

https://doi.org/10.1016/j.ress.2008.03.010

Coherent Market Insights. (2023, November). Paper cone market size and share analysis - Growth trends

and forecasts (2023-2030). coherentmarketinsights.

https://www.coherentmarketinsights.com/industry-reports/paper-cone-market#:~:text=Global%20Paper%20

Cone%20Market%20is,7.3%25%20during%20the%20forecast%20period

COMATEC. (2023). PAPER CONE. comatec.

https://www.comatec.fr/gb/1569-paper-cone.html

Geddes, D. (2021, April 10). 6 Required Functions of All Product Packaging.

technicalfoamservices.

https://technicalfoamservices.co.uk/blog/6-required-functions-of-all-product-packaging/

Khan Academy. (n.d.). Lagrange multipliers, introduction. Khanacademy.

https://www.khanacademy.org/math/multivariable-calculus/applications-of-multivariable-derivatives/constrai

ned-optimization/a/lagrange-multipliers-single-constraint

KORPACK. (2024, July 11). The Rising Costs of Corrugated Packaging in the U.S.:

An In-Depth Analysis. korpack.

https://korpack.com/the-rising-costs-of-corrugated-packaging-in-the-u-s-an-in-depth-analysis/

math.washington. (n.d.). Sylvester's Criterion. math.washington.

https://sites.math.washington.edu/~morrow/334_19/sylvester%20positive%20definite.pdf

24
MARINELAND. (n.d.). Standard Black Retangular Aquarium. marineland.

https://www.marineland.com/products/aquariums/standard-black-retangular-aquarium.aspx

Tetra Pak. (n.d.). Tetra Pak in figures. tetrapak.

https://www.tetrapak.com/ about-tetra-pak/who-we-are/facts-figures

The Coca-Cola Company. (n.d.). How many drinks does The Coca‑Cola Company sell worldwide each day?

Coca-cola.

https://www.coca-cola.com/ng/en/about-us/faq/how-many-drinks-does-the-coca-cola-company-sell-worldwi

de-each-d

25