Professional Documents
Culture Documents
The Maths Behind The Gherkin
The Maths Behind The Gherkin
(My Decorative)
1
Introduction
as the Gherkin. The Gherkin was a building that had enthralled me from a
young age and the possibility of being able to tie it back to maths
intrigued me.
ellipse. I saw that I could use this to model the Gherkin’s surface area
using a tool called surface areas of revolution, which finds the surface area
of objects with a circular cross-section, like the Gherkin. I also realised that I could find the
building’s surface area by using the area of the triangles on its outer edge and adding up their
respective area. This led me to my aim which was to compare these two methods of finding
surface area and to see if, and how, differences might arise between the two. This was especially
exciting since I couldn’t find anybody who had tried to find the surface area of the Gherkin using
its triangles.
However, before I could do this with the Gherkin, I wanted to apply these two methods to
a model that would replicate the Gherkin to help me grasp the concepts of surface areas of
revolution and surface area by the addition of triangle area, and understand how I could apply
these to ellipses.
2
My first step was to explain the concept of surface area of revolution and find an equation
for it.
A surface area of revolution, 𝑆𝐴 𝑟, is the surface area, 𝑆𝐴, of a function that has been rotated 2π
radians about an axis. Using the same principle as with Riemann sums, we can break a rotated
function up into component blocks, in this case, made of cylinders, as shown in Figure 2. The
approximate surface area of this surface area of revolution would be the summation of the
surface areas of the curved space of all the component cylinders, which each have a surface area
of
𝑆𝐴 = 2π × 𝑦 × ∆𝑥,
like in figure 2, where the surface area of revolution of a function of 𝑥 is divided into cylinders.
3
However, as we want to find the surface area of revolution and not the surface area of component
cylinders, there won’t just be a change in 𝑥 but also a change in 𝑦, as we’ll be considering the
length along a curve, 𝑠, or curve length, rather than the straight length along the curved surface
2 2
𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑡𝑤𝑜 𝑝𝑜𝑖𝑛𝑡𝑠 = ∆𝑥 + ∆𝑦 .
As the changes in 𝑥 and 𝑦 tend towards 0, we can represent change in 𝑥 by 𝑑𝑥 and change in 𝑦
2 2
𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑝𝑜𝑖𝑛𝑡𝑠 𝑓𝑜𝑟 𝑛𝑢𝑑𝑔𝑒𝑠 𝑖𝑛 𝑥 𝑎𝑛𝑑 𝑦 = 𝑑𝑥 + 𝑑𝑦
2 2
𝑑𝑥 +𝑑𝑦 2
⇒ 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑝𝑜𝑖𝑛𝑡𝑠 𝑓𝑜𝑟 𝑛𝑢𝑑𝑔𝑒𝑠 𝑖𝑛 𝑥 𝑎𝑛𝑑 𝑦 = 2 × 𝑑𝑥
𝑑𝑥
𝑑𝑦 2
⇒ 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑝𝑜𝑖𝑛𝑡𝑠 𝑓𝑜𝑟 𝑛𝑢𝑑𝑔𝑒𝑠 𝑖𝑛 𝑥 𝑎𝑛𝑑 𝑦 = 1 + ( 𝑑𝑥 ) × 𝑑𝑥.
As I’ve found the equation for the distance between two points for nudges in 𝑥 and 𝑦, as 𝑥 and 𝑦
tend towards 0, we can use the same principle as a Riemann sum to integrate this to find the
𝑏
𝑑𝑦 2
𝑠 =∫ 1 + ( 𝑑𝑥 ) 𝑑𝑥. Equation 1
𝑎
We can see how the length along a curve works visually in figure 3.
4
Now that I have the equation for the length along a curve, I can substitute this into the
place of change in 𝑥 in the formula I had for the surface area of the component cylinders and can
integrate this to find the equation for surface area of revolution. Therefore,
𝑏
𝑑𝑦 2
𝑆𝐴 𝑟
= ∫ 2π × 𝑦 1 + ( 𝑑𝑥 ) 𝑑𝑥
𝑎
𝑏
𝑑𝑦 2
⇒ 𝑆𝐴 𝑟
= 2π ∫ 𝑦 1 + ( 𝑑𝑥 ) 𝑑𝑥. Equation 2
𝑎
This can be seen in Figure 4 where we can see how we have moved from the cylinders of figure
Now that I have found the equation for a surface area of revolution, I could apply this to
an ellipse and see how it would work on a model similar to the Gherkin.
5
An ellipse takes on the form of an oval with an equation which takes the form of
2 2
(𝑦−𝑘) (𝑥−𝑚)
2 + 𝑞
= 1.
𝑝
In an ellipse, the point (𝑚, 𝑘) marks the centre of the ellipse, the constant 𝑝 determines the
distance between the centre of the ellipse and the points on the ellipse whose tangents are parallel
to the x-axis, which is called the semi-minor axis, whilst the constant 𝑞 determines the distance
between the centre and the points on the ellipse whose tangents are parallel to the y-axis, which
is called the semi-major axis. These roles can also be seen visually in figure 5.
We can see from figure 5 that, in more simple terms, 𝑚 and 𝑘 determine the ellipse’s position on
the x-y plane and 𝑝 and 𝑞 determine the ellipse’s size and shape.
Using this general form, we can make a hypothetical model ellipse, like
2 2
𝑦 (𝑥−10)
2 + 2 = 1, Equation 3
7 25
6
for example. This ellipse has a centre of (10,0), a semi-minor axis of 7 units and a semi-major
axis of 25 units. To show equation 3 visually I plotted its graph on GeoGebra, which can be seen
in Figure 6.
I came up with equation 3 because I thought its shape was similar to the Gherkin’s and it
would enable me to look at how the two methods of finding surface area that I’m exploring
would apply to a model similar to the Gherkin. I also decided to flip the model on its side, which
I will do for the Gherkin as well, meaning that the height from the bottom, 𝑥, would be along the
x-axis and half the width of the model at a given height, 𝑦, would be on the y-axis. I did this
because I found my surface area of revolution equation for rotations about the x-axis.
I also decided to restrict the equation of the model to the first quadrant, which I would do
with the Gherkin as well. Restricting the equation to the first quadrant meant that when I applied
the surface area of revolution to my model and the Gherkin, it would be clear that I would only
find the surface area of revolution of the structures above the ground, rather than find the surface
Now that I had an equation for my model, I wanted to restrict it to the first quadrant by
2 2
𝑦 (𝑥−10)
2 + 2 =1
7 25
2
(𝑥−10)
⇒𝑦 = 7 1 − 2
25
7 2 2
⇒𝑦 = 25
25 − (𝑥 − 10) , 𝑥 ≥ 0. Equation 4
Equation 4 is the same curve as equation 3, apart from it being restricted to the first quadrant.
This clearly shows the part of the model I’m applying a surface area of revolution to. Equation 4
Before I could find the surface area of revolution of my model using equation 2, I first
had to find the derivative 𝑦 in equation 4. I did this using the chain rule.
7 2 2
𝐹𝑜𝑟 𝑦 = 25
25 − (𝑥 − 10)
𝑑𝑦 7 1 𝑑 2 2
⇒ 𝑑𝑥
= 25
( 2 2
)( 𝑑𝑥 (25 − (𝑥 − 10) ))
2 25 −(𝑥−10)
𝑑𝑦 7 1
⇒ 𝑑𝑥
= 25
( 2 2
)(− 2(𝑥 − 10))
2 25 −(𝑥−10)
𝑑𝑦 70−7𝑥
⇒ 𝑑𝑥
= 2 2
. Equation 4.5
25 25 −(𝑥−10)
8
𝑑𝑦
Now that I had found 𝑑𝑥
of equation 4 in equation 4.5, I could find the surface area of revolution
𝑑𝑦
of my model. I did this by substituting 𝑦 in equation 4 and 𝑑𝑥
in equation 4.5 into equation 2. I
35
7 2 2 70−7𝑥 2
𝑆𝐴 𝑟
= 2π × ∫( 25 25 − (𝑥 − 10) 1 +( 2 2
) ) 𝑑𝑥
0 25 25 −(𝑥−10)
2
⇒ 𝑆𝐴 𝑟
= 1320 𝑢𝑛𝑖𝑡𝑠 𝑡𝑜 𝑡ℎ𝑒 𝑛𝑒𝑎𝑟𝑒𝑠𝑡 𝑤ℎ𝑜𝑙𝑒 𝑓𝑖𝑔𝑢𝑟𝑒.
I decided to give my final answers in whole figures, as it made sense as a degree of precision for
surface areas on the scale of the Gherkin that I would be dealing with. I’ll show my work and
intermediate calculation to 1 decimal place as that was the highest degree of precision I could
read off the architect’s models of the Gherkin I used. The exception to this is integer numbers,
which will be shown as a whole number. I will also keep track of all intermediate calculations to
the furthest degree of precision that my calculator allowed to ensure all my values are correct
when I’m showing them. For equations which require more than 1 decimal place of precision,
I’ll show the full equation, before displaying it to 1 decimal place. Now that I had found the
surface area of revolution for my model, I could move onto finding its area using triangles.
I wanted to find the surface area by addition of triangle area, 𝑆𝐴 𝑡, of my model so I could see
how the triangles on the Gherkin, which you can see in figure 1, might be used to find the
Gherkin’s surface area. To do this I had to divide my model up into triangles, whose area I would
then add together to find the total surface area. I would then be able to compare the two values
which I would call tiers, 𝑡 , where n is the ordered number of the tier from 0 to 6, with 0 being
𝑛
the number for the tier at the bottom of the model and 6 being the tier for the highest tier. I show
It’s important to note that when I mean tier, I don’t mean the intervals that are divided up on
figure 8, but rather the lines that divide my model up, with each tier being labelled to its right. In
a certain sense, it’s similar to the distinction between a border and a country, a tier would be
analogous to a border whilst the interval between them would be analogous to a country. Next, I
number on which the base, 𝑏 , where n is once again the tier number, of the triangle lies. This is
𝑛
10
shown for tiers 0 to 2 in figure 9, where I unravel the tiers and show how I divide the intervals
Figure 9 shows how 14 triangles have their bases on each tier, apart from tier 0 where there are
only 7. This is because on each tier there are 7 triangles that go upwards out of the tier and
downwards below the tier, as you can see if you look at the triangles whose bases are on tier 1 in
figure 9. However, for tier 0 there are no triangles that come out below and so only the 7
triangles which go up out of tier 0 would use tier 0 for their base, as you can see in figure 9.
I decided on 7 tiers and 14 triangles when I was dividing my model up as I thought these
numbers would be similar to those on the Gherkin, accounting for the different size of my model,
and would therefore enable me to draw similarities between the results for my model and the
Gherkin.
11
Now I needed to decide how to find the height from the bottom at which each tier was
of a specific tier from the height from the bottom in general. As I looked at models and photos of
the Gherkin, it seemed that the height of each triangle was constant, therefore, I decided to keep
the height of each triangle, ℎ, as a constant for all the tiers. This meant that the triangles whose
bases were on the same tier would be congruent since the height was constant and their bases
would be the same length. I saw that the height of each triangle was roughly the same as the
length along the curve of an interval between two tiers for my model (and therefore it would be
for the Gherkin as well). If we look at figure 10, we can see that the length along the curve for
the whole of my model, can be divided by the number of intervals to find the triangle height.
12
Seeing this, I could then find the constant triangle height and use it to find the height
from the bottom at which each tier was situated. The height of a triangle would be a seventh of
the length along the curve from the bottom to the top of my model. The length along the curved
𝑑𝑦
outer edge could be found by substituting the 𝑑𝑥
of equation 4.5 into equation 1, giving me
35
70−7𝑥 2
𝑠=∫ 1 + ( 2 2
) 𝑑𝑥
0 25 25 −(𝑥−10)
𝑠 = 37. 2 𝑢𝑛𝑖𝑡𝑠
𝑠
ℎ= 7
37.2
⇒ℎ = 7
⇒ ℎ = 5. 3 𝑢𝑛𝑖𝑡𝑠.
Now I had the constant height of each triangle, meaning that I had the curve length that separated
each tier, as you can see in figure 10. Therefore, I saw that the length along the curve between
the bottom and any given tier would be the height of a triangle multiplied by that tier’s number.
Since I knew the place of the bottom tier at 0 units from the bottom I could use the CAS function
on my calculator to solve for the height of the tier I wanted to find by using equation 1. This
𝑥 𝑛
70−7𝑥 2
∫ 1 + ( 2 2
) 𝑑𝑥 = 𝑛 × 5. 3.
0 25 25 −(𝑥−10)
For example, if I wanted to find the height at which tier 3 was situated then this would give me
𝑥 3
70−7𝑥 2
∫ 1 + ( 2 2
) 𝑑𝑥 = 3 × 5. 3
0 25 25 −(𝑥−10)
13
⇒𝑥 3
= 15. 9 𝑢𝑛𝑖𝑡𝑠.
This means that tier 3 has a height of 15.9 units from the bottom of the model.
I could now use the information from this formula to find the height from the bottom at
which each tier was situated. I could then find the dimensions of all the triangles and could begin
divide by 7 to find the base of a triangle in a given tier as I show in figure 11, where you can see
that the length of the circumference of a tier is 7 times the length of the base of a triangle by
looking at how my model looks when unravelled and how the bases of a tier relate to a tier’s
To find an expression for the circumference of the model at a given tier, I substituted 𝑦for
equation 4 into the place of the radius for the equation of the circumference of a circle, and I
14
changed 𝑥 to 𝑥 𝑛
to accord for finding the circumference at a given tier, for the reasons I’ve
7 2 2
𝐶 𝑛
= 2π × 25
25 − (𝑥 𝑛
− 10)
14π 2 2
⇒𝐶 𝑛
= 25
25 − (𝑥 𝑛
− 10) .
14π 2 2
𝐶 3
= 25
25 − (𝑥 3
− 10)
14π 2 2
⇒𝐶 3
= 25
25 − (15. 9 − 10)
⇒𝐶 3
= 42. 7 𝑢𝑛𝑖𝑡𝑠.
1
𝑏 𝑛
= 7
× 𝐶 𝑛
1 14π 2 2
⇒𝑏 𝑛
= 7
× 25
25 − (𝑥 𝑛
− 10)
2π 2 2
⇒𝑏 𝑛
= 25
25 − (𝑥 𝑛
− 10) .
2π 2 2
⇒𝑏 3
= 25
25 − (𝑥 3
− 10)
2π 2 2
⇒𝑏 3
= 25
25 − (15. 9 − 10)
⇒𝑏 3
= 6. 1 𝑢𝑛𝑖𝑡𝑠.
Now I had found the dimensions of the triangles, as you can see in figure 12, where the
1
𝐴 𝑛
= 2
× 𝑏 𝑛
× ℎ,
substituting for 𝑏 𝑛
and ℎ,
1 2π 2 2
⇒𝐴 𝑛
= 2
× 25
25 − (𝑥 𝑛
− 10) × 5. 3
π 2 2
⇒𝐴 𝑛
= 5. 3 × 25
25 − (𝑥 𝑛
− 10)
π 2 2
𝐴 3
= 5. 3 × 25
25 − (𝑥 3
− 10)
π 2 2
⇒𝐴 3
= 5. 3 × 25
25 − (15. 9 − 10)
2
⇒𝐴 3
= 16. 2 𝑢𝑛𝑖𝑡𝑠 .
16
Having found the way to get the area of all the triangles, I needed to add up their area to
of my model by addition of triangle area. To find the total area of triangles in a tier, I needed to
multiply the number of triangles in a tier by the area of a triangle in that tier. Since not all tiers
have the same number of triangles, there wouldn’t be a universal expression. For the tiers from
∑𝐴 𝑛
= 14 × 𝐴 𝑛
π 2 2
⇒ ∑𝐴 𝑛
= 14 × 5. 3 × 25
25 − (𝑥 𝑛
− 10)
14π 2 2
⇒ ∑𝐴 𝑛
= 5. 3 × 25
25 − (𝑥 𝑛
− 10) , 𝑓𝑜𝑟 1 ≤ 𝑛 ≤ 6.
π 2 2
⇒ ∑𝐴 0
= 7 × 5. 3 × 25
25 − (𝑥 0
− 10)
7π 2 2
⇒ ∑𝐴 0
= 5. 3 × 25
25 − (𝑥 0
− 10) .
Therefore to find the total area of the triangles who have their base on tier 3, I would do
14π 2 2
∑𝐴 3
= 5. 3 × 25
25 − (𝑥 3
− 10) × 5. 3
14π 2 2
⇒ ∑𝐴 3
= 5. 3 × 25
25 − (15. 9 − 10) × 5. 3
2
⇒ ∑𝐴 3
= 227. 0 𝑢𝑛𝑖𝑡𝑠 .
17
I applied all of the above calculations to all the tiers as detailed, and the results are shown
in table 1.
Table 1 - Data Used to Calculate the Surface Area by Addition of Triangle Area for my Model
Tier, Height of Circumference Base of Area of a Number of Total Area
𝑡 𝑛 Tier From of Tier, 𝐶 𝑛 Triangle, Single Triangles for Triangles
Bottom, (in units) 𝑏 𝑛 Triangle, whose base in a Tier,
𝑥 𝑛 (in units) 𝐴 𝑛 is on Tier, 𝑇𝐴 𝑛
(in units) (in 𝑡 𝑛 2
(in𝑢𝑛𝑖𝑡𝑠 )
2
𝑢𝑛𝑖𝑡𝑠 )
I then added up all the areas to find the surface area of my model by addition of triangle
2
area and got a surface area of 1302 𝑢𝑛𝑖𝑡𝑠 . Having found this, I could compare it with my value
I was very happy to see how close the two values for surface area were to each other. A
2
difference of only 18 𝑢𝑛𝑖𝑡𝑠 was much closer than I thought they might be. As I thought about
why these values had a difference at all, I realised that finding the surface area by addition of
triangle area considered my model as being made up entirely of triangles, like with the Gherkin.
However, a surface area of revolution considered my model as being a smooth and curved
revolution. As I further reflected on this, I realised that being made up of triangles would only
consider a flat 2-dimensional surface for the components to find surface area. On the other hand,
a surface area of revolution would add a curved element to the model that would bring in a
3-dimensional surface to consider for the surface. This should mean a greater surface area than
for the surface area by addition of triangle area. Some of this difference could also be explained
by the discrepancy I mentioned earlier in using the length along the curve of the outer edge of
my model for the triangle height. This same discrepancy would arise when I worked with the
Gherkin.
19
I wanted to test the hypothesis about the different dimensions being a reason for the different
values for surface area, so I decided to build my model divided into triangles using paper and
spent a day putting this together, using a ratio of 1 𝑢𝑛𝑖𝑡: 1 𝑐𝑒𝑛𝑡𝑖𝑚𝑒𝑡𝑟𝑒, to observe how this
would turn out visually. The results of this can be seen in figures 13 and 14, shown above. I was
very pleased to see how this turned out and thought it showed a bulge and curve that was similar
to the Gherkin and typical of an ellipse. The curve in my model can be seen in particular in
20
figure 15, where I show the interval between tiers and 5 unravelled. I was also very happy with
how the dimensions turned out as I measured it and its height was within a centimetre of the
expected height of 35 centimetres. Whilst, I thought this looked similar to equations 3 and 4
which was supposed to model its outer edge, I wanted to verify this by putting my model up
I was happy to see that the outer edge of my built model was quite close to the equation
that modelled it. Despite using the exact same dimensions that I outlined earlier which were
based on the measurements I took using equations 3 and 4, the model when divided into triangles
was thinner at the top and bottom of it, as you can see from figure 16. This suggested that the
surface area of my built model would be less than the surface area of revolution of equation 4.
This would confirm my aforementioned belief that finding the surface area by addition of
triangle area would lead to a smaller value than surface area of revolution. I could then
hypothesise that the same would happen for the Gherkin but with a much larger difference
between the two since the dimensions modelling the Gherkin were much greater than those of
my model.
21
Now, I could move to find the surface areas of the Gherkin by surface area of revolution
Having shown how the two methods of finding surface area worked with a hypothetical model, I
was eager to apply them to my real-life example of the Gherkin and see if I could draw similar
conclusions.
To model the Gherkin’s outer edge, I used a to-scale architect’s model of the Gherkin
(Skyscrapercenter), which can be seen in figures 18 and 19, and I rotated it around 2π radians, to
set it along the x-axis, for the same reasons I explained earlier with my model. Using the
Gherkin’s dimensions of being 179.5 metres tall from the ground to the top, 56.5 metres wide
across at its widest point (Wikiarquitectura) and 49 metres wide across at the ground
(Wikiarquitectura), I resized the architect’s model using a ratio of 1 unit: 1 metre to fit these
dimensions and then found the equation that best fit its outer edge, giving me
2 2
𝑦 (𝑥−61.85)
2 + 2 = 1, Equation 5
28.75 117.65
where 𝑥 is once again the height from the ground in metres and 𝑦 is again half the building’s
width at a given height, 𝑥. Since I placed the building along the x-axis, 𝑦 had to be equal to 0 at
the centre of the building. Since half of the maximum width of 56 metres is 28.75 metres, this
2
𝑦
must be the length of the semi-minor axis. These two facts put together gave the 2
28.75
component of equation 5. Then I knew that the length of the semi-major axis and the
displacement of the centre from the origin had to add up to 179.5. So I adjusted these two until
they fit in with the scaled out architect’s model of the Gherkin. This gave me a displacement of
22
61.85 metres along the x-axis from the origin and a semi-major axis of 117.65 metres. These
2
(𝑥−61.85)
factors led to the 2 component of equation 5.
117.65
Having found all the components of equation 5, I verified it by checking the width of this
ellipse at 𝑥 = 0 𝑚𝑒𝑡𝑟𝑒𝑠 (i.e. the width of the Gherkin at its bottom) by looking at the graph
visually and I got a width of 48.92 m. This was extremely close to the actual value of 49 metres I
mentioned above and pleased me as it showed that my equation was very accurate. How all of
I then plotted equation 5 against the architect’s model of the Gherkin to be able to see visually
how this would turn out. This can be seen in figure 18 below.
23
I was really happy with how well equation 5 approximated the edge of the Gherkin. I
noticed, however, that there was a small discrepancy between the building’s outer edge and the
equation towards the top of the building. This was much like my model where the top had the
same sort of discrepancy between my built model and equations 3 and 4 upon which it was
based, however, the base of the Gherkin was modelled much more accurately by equation 5 than
my built model was by equations 3 and 4. By looking at figure 18 I determined visually that
equation 6 was accurate enough that any discrepancies it had in modelling the Gherkin’s outer
Like with my earlier model, I restricted equation 5 to the first quadrant by solving for 𝑦
2
(𝑥−61.9)
⇒ 𝑦 = 28. 8 1 − 2
117.7
28.8 2 2
⇒𝑦 = 117.7
117. 7 − (𝑥 − 61. 9) , 𝑥 ≥ 0. Equation 6
Equation 6 and how it models the Gherkin can be seen visually in figure 19 below.
24
Like before I needed to find the derivative of equation 6 before I could find the Gherkin’s surface
area of revolution. I once again used the chain rule, which meant that
28.8 2 2
𝑓𝑜𝑟 𝑦 = 117.7
117. 7 − (𝑥 − 61. 9)
𝑑𝑦 28.8 1 𝑑 2 2
⇒ 𝑑𝑥
= 117.7
( 2 2
)( 𝑑𝑥 (117. 7 − (𝑥 − 61. 9) )
2 117.7 −(𝑥−61.9)
𝑑𝑦 28.8 1
⇒ 𝑑𝑥
= 117.7
( 2 2
)(− 2(𝑥 − 61. 9))
2 117.7 −(𝑥−61.9)
𝑑𝑦 1778.2−28.8𝑥
⇒ 𝑑𝑥
= 2 2
. Equation 6.5
117.7 117.7 −(𝑥−61.9)
Before I found the surface area of revolution, I decided to only consider the building’s
surface from 7.1 metres up. I decided this because as you can see from the architect’s model in
figures 18 and 19, the building’s first row of triangles, which goes from the bottom to 7.1 metres
high, is incomplete as part of the top is missing from the triangles going into the ground. This
Having made this decision I then found the surface area of revolution of the Gherkin
𝑑𝑦
from 7.1 metres to its top by substituting 𝑦 in equation 6 and 𝑑𝑥
from equation 6.5 into equation
179.5
28.8 2 2 1778.2−28.8𝑥 2
𝑆𝐴 𝑟
= 2π × ∫ ( 117.7 117. 7 − (𝑥 − 61. 9) 1+( 2 2
) ) 𝑑𝑥
7.1 117.7 117.7 −(𝑥−61.9)
2
⇒ 𝑆𝐴 𝑟
= 26648 𝑚𝑒𝑡𝑟𝑒𝑠
Now that I had the surface area of revolution for the Gherkin from 7.1 metres to the top, I
could find its surface area by addition of triangle area for that region of the building.
25
I used the same method to find the surface area of by addition of triangle area, as I did with my
Model. The big difference between how I would apply this to the Gherkin and my model is that,
as I’ve already mentioned, I would begin counting the tiers of the Gherkin from the 7.1 metres
up the building. I would also obviously be using a different number of tiers and triangles, than I
used with my model, as well as a different equation (equation 6 instead of equation 4) in the
calculation.
I observed visually that from 7.1 metres up to the top of the Gherkin, there were 21 tiers
and 36 triangles in the intervals between each tier. I would number the tiers from 0 to 20, and I
could also see that each tier would have 36 triangles using it for its base, apart from the bottom
tier, which would have 18, for the same reasons as I discussed with my model.
To find the height of each triangle, and thus the curve length along the Gherkin’s outer
edge between tiers, instead of finding the curve length from the bottom to the top and divide by, I
needed to find the curve length from 7.1 metres up to the top. Using this knowledge, I went
𝑑𝑦
through the same process as I did before, substituting 𝑑𝑥
from equation 6.5 into equation 1,
which gave me
179.5
1778.2−28.8𝑥 2
𝑠= ∫ 1 +( 2 2
) 𝑑𝑥
7.1 117.7 117.7 −(𝑥−61.9)
⇒ 𝑠 = 180. 7 𝑚𝑒𝑡𝑟𝑒𝑠.
Like with my model, I then divided this by the number of tiers (21, as already mentioned) to find
the triangle height, and thus the curve length between tiers, giving me
ℎ = 8. 6 𝑚𝑒𝑡𝑟𝑒𝑠.
26
Using the same principle as with my model, I found the height from the bottom at which each
𝑥 𝑛
1778.2−28.8𝑥 2
∫ 1+ ( 2 2
) 𝑑𝑥 = 𝑛 × 8. 6,
7.1 117.7 117.7 −(𝑥−61.9)
where I began at 7.1 metres up rather than from the figurative ground, as I’ve already discussed.
I then went through the exact same process as I did with my model to find the surface
area by addition of triangle area, except I substituted equation 6 where I had previously
substituted equation 4, and of course with the triangle height and number of triangles for the
Gherkin. All of the equations I used are shown below, and the values I found using them can be
28.8 2 2
𝐶 𝑛
= 2π × 117.7
117. 7 − (𝑥 𝑛
− 61. 9) .
2π 28.8 2 2
𝑏 𝑛
= 18
× 117.7
117. 7 − (𝑥 𝑛
− 61. 9) .
π 28.8 2 2
𝐴 𝑛
= 8. 6 × 18
× 117.7
117. 7 − (𝑥 𝑛
− 61. 9) .
π 28.8 2 2
∑𝐴 𝑛
= 36 × 8. 6 × 18
× 117.7
117. 7 − (𝑥 𝑛
− 61. 9)
28.8 2 2
⇒ ∑𝐴 𝑛
= 8. 6 × 2π × 117.7
117. 7 − (𝑥 𝑛
− 61. 9) , 𝑓𝑜𝑟 1 ≤ 𝑛 ≤ 20.
27
π 28.8 2 2
∑𝐴 0
= 18 × 8. 6 × 18
× 117.7
117. 7 − (𝑥 0
− 61. 9)
28.8 2 2
⇒ ∑𝐴 0
= 8. 6 × π × 117.7
117. 7 − (𝑥 0
− 61. 9) .
Once again, I then added up all the areas for the triangles of each tier to find the total
Table 2 -Data Used to Calculate the Surface Area by Addition of Triangle Area of the Gherkin
Tier, Height of Tier Circumference of Base of Area of a Single Number of Total Area for Triangles in a
𝑡 𝑛 From Bottom, Tier, 𝐶 𝑛 Triangle, Triangle, 𝐴 𝑛 Triangles in Tier, 2
𝑥 𝑛 (in metres) 𝑏 𝑛 (in 𝑡 𝑛 Tier, ∑ 𝐴 (in𝑚𝑒𝑡𝑟𝑒𝑠 )
(in metres) (in 𝑚𝑒𝑡𝑟𝑒𝑠 )
2 𝑛
metres)
𝑡 18
0 7.1 159.9 8.9 38.2 688.1
𝑡 36
1 15.6 166.1 9.2 39.7 1429.8
𝑡 36
2 24.2 171.2 9.5 40.9 1473.0
𝑡 36
3 32.8 175.0 9.7 41.8 1506.6
𝑡 36
4 41.4 177.9 9.9 42.5 1531.0
𝑡 36
5 50.0 179.7 10.0 43.0 1546.8
𝑡 36
6 58.6 180.6 10.0 43.2 1554.1
𝑡 36
7 67.2 180.5 10.0 43.1 1553.1
𝑡 36
8 75.8 179.4 10.0 42.9 1543.7
𝑡 36
9 84.4 177.3 9.9 42.4 1525.8
𝑡 36
10 93.0 174.2 9.7 41.6 1499.2
𝑡 36
11 101.6 170.0 9.4 40.6 1463.3
𝑡 36
12 110.2 164.7 9.2 39.4 1417.6
𝑡 36
13 118.7 158.2 8.8 37.8 1361.2
𝑡 36
14 127.2 150.2 8.3 35.9 1292.7
𝑡 36
15 135.7 140.6 7.8 33.6 1210.5
𝑡 36
16 144.1 129.2 7.2 30.9 1111.8
𝑡 36
17 152.4 115.3 6.4 27.6 992.5
𝑡 36
18 160.6 98.3 5.5 23.5 845.7
𝑡 36
19 168.4 76.5 4.2 18.3 658.0
𝑡 36
20 175.6 46.3 2.6 11.1 398.8
2
26603 𝑚𝑒𝑡𝑟𝑒𝑠 for the Gherkin. Having found the value for the surface area of the Gherkin by
Comparing the Two Methods to Find Surface Area for the Gherkin
I was delighted with these results. Having gone through the same process with my model, I
thought that for the much larger scale Gherkin, the difference between the two values of surface
2
area would be much larger than the 45 𝑚𝑒𝑡𝑟𝑒𝑠 that it was. I very quickly realised that this was
because there are more tiers and triangles in the Gherkin than I used in my model. This meant
that whilst the absolute difference between the two values was greater for the Gherkin, the
percentage difference was much greater for my model. Why this is can be explained with the
If you consider some curve that can be expressed as a function of 𝑥 in between an interval from
any given point A to any given point B, as I did with figure 20, then you can very roughly
approximate the distance along the curve from A to B by the straight distance from A to B. This
is what I did to find the constant height of the triangles in both my model and the Gherkin, where
I used the curve to find the surface area of revolution, whilst I considered straight lines to find
the surface area by addition of triangle area. There will obviously be some discrepancy between
the length of a straight line between the two points and the length along the curve between the
two points. However, when I inserted a midpoint, M, along the curve between A and B and drew
two lines to get from A to B, going through M, I could see visually but still very clearly that the
discrepancy between the sum of the length of these two lines and the length along the curve from
A to B was less than the discrepancy between the straight line from A to B and the length along
the curve from A to B. In applying this to my exploration, this meant that the more tiers and
triangle I broke my surface areas of revolution up into, the less of a discrepancy I would have
between the methods I used to find surface area of revolution and surface area by addition of
triangle area, which explained why the two values that I found for the surface area of the Gherkin
were a lot closer than I expected, simply because the Gherkin has more tiers and triangles than I
These two values for surface area of the Gherkin were slightly further apart from the
2
actual surface area of the Gherkin, which is 24000 𝑚𝑒𝑡𝑟𝑒𝑠 (Pauling) than they were from
each other, especially since this considers the whole building, rather than only from 7.1 metres
up and above, which means they’d be even further apart if the surface area of the Gherkin was
adjusted to find the surface area 7.1 metres, like I did, and above rather than the whole building.
This is almost certainly due to the discrepancy between the Gherkin and the equations 5 and 6,
31
which I used to model it, at the top. Nevertheless, they were still close enough to the Gherkin’s
actual surface area that my belief that any discrepancies between my calculations and the actual
surface area of the Gherkin wouldn’t be excessive was confirmed. In fact, the biggest distortion
from the discrepancy at the top came in the differences between the heights from the ground at
which each tier was situated that I used and their actual position. You can check this by
comparing the values from table 2 with a visual reading from the architect’s model present in
figures 18 and 19. This distortion is almost entirely due to the discrepancy at the top, as I
checked the values whilst only considering the parts of the Gherkin that were best modelled by
equations 5 and 6 and they were all exactly the same as the values on the architect’s model of the
Conclusion
This exploration helped me explore two different ways to find the surface area and understand
how they differed. It showed me that of the two methods I used, surface area of revolution gave a
larger value than surface area by addition of triangle area. Exploring these two methods on a
model before moving to their application on the Gherkin helped me understand how each of
them functioned. Building a paper model was very useful in clearly showing the difference
between these two methods by giving me a visual input that enabled me to best compare them. I
also clearly saw how approximating straight lines with curved lengths would lead to
discrepancies between the two. This also showed me how using more tiers and triangles would
bring the values of surface area by these two methods closer together. This led me to think that if
I used the same principle as a Riemann sum and went towards an infinite number of tiers and
triangles the value for surface area by addition of triangle area would tend towards the value for
32
the surface area of revolution, however, applying this to real life in architecture would be neither
Works Cited
Pauling, Keith. "The Gherkin." thamespathway.com, edited by Owen Pauling, Owen Pauling,
Jan. 2021.
Wikiarquitectura, wikiarquitectura,