You are on page 1of 14

Case Study:

Reviewing a Proposed Modification


for a Fairground Ride

Ethan Holmes
18-11-20 │ Department of Mechanical Engineering, Imperial College London
Abstract

This report reviews the safety impacts of a proposed modification to a rotating bar in a
fairground ride. The nominal stress in the part is calculated. Tensile tests are performed to
determine the Young’s Modulus and yield, fatigue and ultimate stresses using two different
measuring systems. Strain gauges are used to find a Stress Concentration Factor resulting from
the altered geometry of the new part. Results are presented as safety factors before and after the
modification, and a significant decrease in margin is demonstrated. The report advises against
making the modification as proposed and suggests further fatigue analysis.

Nomenclature

E Young’s Modulus, Pa ε Strain


F Force, N σ Stress, Pa
Fc Centripetal Force, N σf Fatigue Stress, Pa
Kt Stress Concentration σUTS Ultimate Tensile Stress,
Factor Pa
L0 Original Length, m σy Yield Stress, Pa
ΔL Extension, m ω Angular Velocity,
Rads-1
m Mass, Kg BV Bridge Voltage, V
r Radius, m GF Gauge Factor
δ Relative uncertainty, % SF Safety Factor
Introduction

Riders of fairground attractions are often subjected to high accelerations and strong
forces. Therefore, it is imperative that a suitable amount of attention is given to ensuring their
safe operation. This report, commissioned by the Health and Safety Executive, assesses the
safety of a proposed modification to a ride consisting of cars, supported by tracks, which are spin
around a central drive shaft. The mild steel bar connecting the shaft and cart is to potentially
have a 25mm diameter hole drilled to facilitate the installation of lights (see figure 1). This
investigation will use a range of laboratory techniques with the aim of determining whether the
modification can be deemed acceptably safe.

Figure 1 Modified part: a) Render b) Drawing


Aims
The greater objective can be broken down into these aims:
• The forces applied to the bar will be found, and the risk of failure through yielding will
be calculated.
• A fatigue analysis will be undertaken to ensure the modification is within acceptable
safety margins for fatigue failure.
• The Young’s Modulus E of the mild steel used in the bar will be determined to validate
our methodology.

Determining Stresses
First the operating stress on the part must be found. It is reasonable to assume the sole
force acting on the bar is centripetal force from the rotation, as the car is supported from below
by the tracks it rides.
This force can be found thus:

𝐹𝑐 = 𝑚𝑟𝜔2

For the unmodified bar, this gives a force of 24.7kN. This gives a stress σ of 30.8MPa
(appendix 1). To find the safety factor of the part, before and after the proposed modification,
this stress will be compared against maximum stresses. Safety factors can be found:

𝜎𝑙𝑖𝑚𝑖𝑡
𝑆𝐹 =
𝜎𝑛𝑜𝑚

Where 𝜎𝑙𝑖𝑚𝑖𝑡 is either the yield stress 𝜎𝑦 or fatigue stress 𝜎𝑓 to give the corresponding
safety factor, and 𝜎𝑛𝑜𝑚 is the operational stress on the part.
Methodology

This investigation will use experimental techniques to find numerical values for 𝜎𝑦 , 𝜎𝑓
and E, and the Stress Concentration Factor (Kt).

Tensile Testing

Measurement of the behavior of the material used for the bar under tensile stress, such as
that experienced in its operation, will allow for the determination of 𝜎𝑦 , 𝜎𝑓 and E for bar’s mild
steel. A sample of mild steel was placed into an Instron Universal Testing Machine and subjected
to tensile stress until failure. The engineering stress:
𝐹
𝜎=
𝐴
As well as strain:
∆𝐿
𝜀=
𝐿0
Are continuously recorded and presented in a stress – strain graph. This should show two
regions of deformation as the stress increases, first a linear section of elastic deformation. The
gradient of this section is the young’s modulus of the material:
𝜎
𝐸= (for elastic deformation)
𝜀

As the sample is further stressed, it will move to plastic deformation, moving through an
unpredictable area of stress-strain behavior known as the Lüders band, (Guy and Hren, 1974).
The lowest stress value recorded here is taken as 𝜎𝑦 , the point at which elastic moves to plastic
deformation. Finally, 𝜎𝑈𝑇𝑆 will be found by taking the greatest stress the sample undergoes
during plastic deformation. This allows for an estimation for 𝜎𝑓 (Forrest, 1962):
2
𝜎𝑓 = 𝜎𝑈𝑇𝑆
3
The sample is placed in the Universal Testing Machine, its diameter taken with a
micrometer at 3 different points, and a mean found. This gives an uncertainty of ±0.005mm per
reading, from the micrometer resolution. The length L0 of the sample between the jaws is then
taken with vernier calipers, with an uncertainty of ±0.05mm. The sample is then gradually
stressed, using the machine’s readout to record extension and a load cell with an uncertainty of
±0.5N to give the corresponding load values.
Figure 2: Tensile testing apparatus

Two tests are performed. One using the machine’s own sensors, and a second within only
the elastic region using a more sensitive extensometer. This is done as the extensometer yields
more precise data but would be damaged over greater strains.
The relevant safety precautions are taken – a guard is used to ensure hands are clear of
the high forces of the machine. Glasses are also worn so fracture particles cannot damage eyes.
Anyone operating the machine must be properly trained.

Determining Stress Concentration Factor

Stress concentration factor, Kt, is the ratio of local stress to the nominal stress where
stress has been concentrated in a non-uniform manner due to the geometry of the part (Peterson
and Pilkey, 2003). It can be defined:
𝜎𝑚𝑎𝑥
𝐾𝑡 =
𝜎𝑛𝑜𝑚
This is determined experimentally by machining an aluminium plate (Kt is material
independent, and aluminium’s higher E gives lower uncertainty) to the proposed dimensions.
The plate is then tensioned, with strain gauges recording strain both around the hole, εmax and
remotely εnom. Since 𝜎 = 𝐸𝜀, the ratio of strains is the same as the stresses.
Figure 3: Kt testing part with strain gauges

The strain gauges output a voltage, proportional to strain. The actual strain is found:
4𝑉
𝜀=
𝐵𝑉 × 𝐺𝐹
Where V is the output voltage, BV the bridge voltage of the measuring circuit and GF the
gauge factor. The data collected in this experiment will be verified twofold: firstly they will be
compared against an FEA (finite element analysis) simulation of the same stress on the part, and
also the 𝜀𝑛𝑜𝑚 value will be used to calculate E with the 𝜎𝑛𝑜𝑚 value, if this figure agrees with the
data book value, it suggests the data are reliable.

Results & Discussion

Tensile Testing and Extensometer

The tensile test produced the below stress-strain curve, which allows us to find values for
𝜎𝑦 and 𝜎𝑈𝑇𝑆 (and thus 𝜎𝑓 ).
500

400

300
σ/MPA

200

100

0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

-100 ε

Figure 4: Tensile test results

350

300
y = 20817x - 213.5

250

200
σ/MPA

150

100

50

0
0 0.005 0.01 0.015 0.02 0.025 0.03
ε

Figure 5: The elastic region of the tensile test


This experiment gave us values 𝝈𝒚 = 327MPa, 𝝈𝑼𝑻𝑺 = 472 MPa and therefore 𝝈𝒇 =
315MP, all ±1.31% (appendix 2). This also gives a failure strain of 0.3473 (or true strain of
0.2981, appendix 6)
The extensometer data for the second tensile test run, in the elastic region, is shown
below:

120

100

80
σ/MPA

60

40

20

0
0 0.00005 0.0001 0.00015 0.0002 0.00025 0.0003 0.00035 0.0004 0.00045 0.0005
ε

Figure 7: Extensometer data plotted for elastic deformation

By computing the gradient of this line, this experiment gives us a value for E of
228.7GPa ±1.9% (appendix 3).

Strain Gauge Data for Kt

The data from the strain gauge test gave us the Kt we needed to understand how the
modification would affect the stress in the part. This gave 𝑲𝒕 = 3.2 ± 8.48% (appendix 4).
Verifying Experimental Results

Both sources of data can be validated: in the case of the tensile test, the value for E was
208GPa, whereas the value from the extensometer testing was higher at 228.7GPa. This is likely
due to error in measuring the gauge length for the tensile test without the extensometer. In
addition, taking only the second, more strictly linear, section of the extensometer data gives a
closer 219.3GPa (appendix 3). These latter values fall in the expected region from data booklets
(ASM International, 1990).
Comparing the 𝑲𝒕 value, and strain values attained experimentally with the FEA
simulation gives this graph:
4.0

3.5 Experimental
Finite Element
3.0

2.5
Stress ratio

2.0

1.5

1.0

0.5

0.0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28
Distance /mm

Figure 8: Comparing FEA and experimental results for strain

Which shows that our experimental data was close to that expected, however slightly
higher at larger distances from the hole, as can be seen with our higher calculated uncertainty in
𝑉𝑛𝑜𝑚 (appendix 4).
Sources of Uncertainty

For the tensile test, the uncertainty in determining the yield and UTS stresses were
relatively low, however some extra uncertainty was introduced by the Lüders bands in
determining the yield stress. The impact of this, however, was minor as the lowest value in the
bands can be taken as an arbitrary value. The fatigue stress value however is based on an
arbitrary approximation. To find it more accurately, a repeated loading test could be carried out,
mimicking operating conditions.
The Kt value had a greater uncertainty. This could be reduced by applying a greater stress
to the material, increasing its strain and thus decreasing the relative uncertainty, or by using a
material with a lower Young’s modulus, for a similar effect.

Conclusion

Considering Safety Factors & Recommendations

To contextualise the above results, they can be shown as safety factors. In the unaltered
part we have:
𝜎𝑦 327 𝜎𝑓 315
𝑆𝐹𝑦 = = = 𝟏𝟎. 𝟔 𝑎𝑛𝑑 𝑆𝐹𝑓 = = = 𝟏𝟎. 𝟐
𝜎𝑛𝑜𝑚 30.8 𝜎𝑛𝑜𝑚 30.8

Using 𝐾𝑡 to find the stress after drilling:

𝜎𝑦 327 𝜎𝑓 315
𝑆𝐹𝑦,ℎ𝑜𝑙𝑒 = = = 𝟑. 𝟑 𝑎𝑛𝑑 𝑆𝐹𝑓,ℎ𝑜𝑙𝑒 = = = 𝟑. 𝟐
𝜎𝑛𝑜𝑚 × 𝐾𝑡 30.8 × 3.2 𝜎𝑛𝑜𝑚 30.8 × 3.2

As can be seen, the safety factor of the part after modification would be significantly
lower than before. From the uncertainties, the worst-case scenario gives an SF of 2.95 (appendix
5), a significant reduction from the original SF of 10.2. Regulations for fairground ride design
specify a SF of “3-5 for more critical components” such as this, leaving this part outside this
nominal range (Atkins, 2007). Arguably the greatest risk identified by this report is that of
fatigue failure: the part is stressed repeatedly as the ride spins and slows, and its fatigue stress (as
estimated here) is lower than its yield strength, implying that it is the first way the part would
fail. In addition, the methodology for acquiring these data didn’t include rigorous fatigue testing
to understand the part’s performance under conditions closer to those experienced in operation.
As a result, this report finds it ill-advised that the modification be made as proposed. It
recommends that any modification isn’t applied, pending the result of a more thorough fatigue
analysis, possibly using repeated load testing, which would yield more relevant data.

This report aimed to judge the safety of a modification to a part of a fairground ride. It
found values for the yield, fatigue and ultimate tensile stress of the mild steel of the part and that
the proposed modifications would serve to increase the local stress in the part by a factor of more
than 3, decreasing the safety factor accordingly, leaving the part more at risk of failure,
especially from fatigue. It concludes that the modification is not advisable before further tests
can be done to better understand the risk of fatigue fracture on the part.

References
• ASM International, 1990. Properties And Selection: Irons, Steels, And High- Performance Alloys.
ASM.
• Atkins, 2007. A Review Of The Design Review Process For Fairground Rides. Health and Safety
Executive.
• Blackman, B. and Al-Nasiri, N., 2020. Material Properties & Phenomena.
• Forrest, P.G., 1962. Fatigue of Metals. Pergamon Press.
• Guy and Hren, 1974. Elements Of Physical Metallurgy. 1st ed. Addison Wesley.
• Peterson, R. and Pilkey, W., 2003. Peterson's Stress Concentration Factors. Norwich, NY:
Knovel.

Appendices

(1) 𝐹𝑐 = 𝑚𝑟𝜔2 = 50 × 5 × 𝜋 2 = 2467𝑁


𝐹 𝐹 𝑟𝑎𝑛𝑔𝑒
(2) Stress uncertainty: 𝜎 = 𝐴 = 𝑑 2 ∴ 𝛿𝑠𝑡𝑟𝑒𝑠𝑠 = 2𝛿𝑑 + 𝛿𝐹 . 𝛿𝑑 = ± 2×𝑚𝑒𝑎𝑛 and 𝛿𝐹 = 1% .
( ) 𝜋
2
This gives 𝛿𝑑 = 0.31% and various 𝛿𝐹 , thus 𝛿𝜎 = 1.31%
Extensometer data - 2nd half
120

100 y = 219272x + 1.4766


80

60

40

20

0
0 0.0001 0.0002 0.0003 0.0004 0.0005
(3)
𝜎
Uncertainty: 𝐸 = plotting max. σ against min. ε for the maximum possible E:
𝜀

120

100 y = 231536x - 1.5233

80

60

40

20

0
0 0.00005 0.0001 0.00015 0.0002 0.00025 0.0003 0.00035 0.0004 0.00045 0.0005

And the opposite:


120

100 y = 223040x - 1.512

80

60

40

20

0
0 0.00005 0.0001 0.00015 0.0002 0.00025 0.0003 0.00035 0.0004 0.00045 0.0005

𝑟𝑎𝑛𝑔𝑒
Gives 𝛿𝐸 = 2×𝑚𝑒𝑎𝑛 = 1.9%

𝑉
(4) Uncertainty in 𝐾𝑡 can be found: 𝐾𝑡 = 𝑉𝑚𝑎𝑥 , 𝛿𝐾𝑡 = 𝛿𝑉𝑚𝑎𝑥 + 𝛿𝑉𝑚𝑖𝑛 . Taking uncertainty as
𝑛𝑜𝑚
𝑟𝑎𝑛𝑔𝑒
±2×𝑚𝑒𝑎𝑛: 𝛿𝐾𝑡 = 0.76% + 7.72% = 8.48%, note 𝑉𝑚𝑖𝑛 has a greater uncertainty.

(5) Worst case 𝐾𝑡 of 3.47 (see (3)) in 𝑆𝐹𝑓,ℎ𝑜𝑙𝑒 gives 2.95

(6) True strain, which takes into account deformation of the original length and is therefore
smaller, is found with ln(1+ε) (Blackman and Al-Nasiri, 2020)

You might also like