MACE43001/MACE 61057: Structural Integrity Lecturers: Dr. KB Katnam & Dr.

AP Jivkov

COURSEWORK SOLUTION & FEEDBACK

DESIGN OF PENSTOCK PIPE FOR A HYDROELECTRIC PUMPED STORAGE STATION
1. OVERVIEW

In this document, you will find a solution strategy for the design task given to you. As a set
of individual parameters are assigned to each student, the solution presented here is for a
general case which can easily be used to compare with your own design approach submitted
for coursework assessment.

2. PUMPED STORAGE POWER STATION: DESIGN OF PENSTOCK

By now, you all know what we mean by a pumped storage power station. The task is to
design a penstock, which is a long pipe assembled with circumferential and longitudinal
welds. The design requirements are:

(a) A 50 year service life considering a given maintenance and repair plan (i.e. non-
destructive inspection every 5 years, and structural repair if crack sizes > 15 mm are
detected);
(b) A water hammer condition (i.e. a maximum service load which is greater than the
loads experienced during regular service cyclic conditions);
(c) Cyclic service conditions (i.e. fatigue crack growth; 7500 cycles per year).

3. DESIGN PROBLEM: OPTIONS

As a design engineer you have a few options to come up with a reliable design which ensures
structural integrity. First, note that you have a specific set of fixed parameters (i.e.
constraints) for the pipe, i.e. internal diameter, internal cyclic pressure, water hammer
pressure, and an initial defect size. Next, note that you have some flexibility via a set of
material and geometric parameters, i.e. either material A or material B, and pipe wall
thickness. Moreover, you have two additional parameters: (a) maximum allowable stress
level (i.e. applied stresses must be ≤ 60% of material yield strength) and (b) internal residual
stresses near longitudinal and circumferential welds. From a design viewpoint, when all the
fixed parameters are carefully accommodated in your analysis, you can meet the design
requirements only by selecting a suitable combination of the two flexible parameters (i.e.
wall thickness and material type).

4. DESIGN STRATEGY: SOME IMPORTANT QUESTIONS

The task is to apply the concepts of ‘linear elastic fracture mechanics’, ‘fatigue crack
growth’, ‘plastic collapse’ and ‘failure assessment diagram’ in combination with the concept
of ‘damage tolerance’ (i.e. considering the maintenance and repair plan).

The following questions are important to address:

(a) What is ‘the critical crack shape and size’?

(b) Should ‘the conditions for static failure’ be assessed first, and then check for ‘the
conditions for fatigue failure’?

Page 1 of 10
MACE43001/MACE 61057: Structural Integrity Lecturers: Dr. KB Katnam & Dr. AP Jivkov

(c) Should ‘the conditions for fatigue failure’ be assessed first, and then check for ‘the
conditions for static failure’?
(d) How can ‘the critical crack size for repair’ (i.e. repairable crack size of at least 15 mm)
be accommodated in ‘the failure assessment’ to ensure detection of such cracks (i.e.
failure should not happen with cracks less than 15 mm) and thus the structural
integrity of the penstock?
(e) Should ‘the residual stresses’ present in welded regions be accounted for in ‘the failure
assessment’?

Overall, to ensure structural integrity, you need to identify critical conditions and then
consider the associated critical parameters in your analysis to come up with a reliable design.

Let us try to answer question (a). For example, cracks can be critical when their orientation
leads to a mode-I type loading condition. For the current design task, as pipes with internal
pressure introduce both longitudinal and hoop stresses (where the magnitude of hoop
stresses is twice that of the longitudinal stresses), a crack in the longitudinal welds is critical
when compared with a crack (with the same size) in the circumferential welds. Following a
similar argument, you can say that the mode-I stress intensity for a crack (with a given size;
here say the depth along the wall thickness direction) which has a semi-elliptical shape is less
critical than a crack (with the same depth along the wall thickness) which is very long along
the longitudinal direction. In short, you can see that the stress intensity factor for an edge
crack in a pipe with internal pressure is critical compared with that of a semi-elliptical crack.
So, for your design, you can assume that you have an initial crack in the pipe which acts like
an edge crack under mode-I loading (here the loading is introduced due to hoop stresses).
Comparing the internal diameter with the wall thickness, you can neglect the curvature
effects and simplify your analysis by assuming it as an edge crack in a finite plate (where
your crack length is the length of the edge crack and the wall thickness is the plate width).

Let us now think about questions (b), (c) and (d). As the task is to design the pipe for a
service life of 50 years, any crack which is present initially should not grow and become a
critical crack during service (i.e. before it can be detected and repaired). As a design
engineer, this is where you play a role when selecting ‘materials’ and ‘geometries’. In this
case, as you have a maintenance and repair plan (i.e. NDT inspections every 5 years, and
repairs for cracks > 15 mm), you need to ensure that the pipe will survive cracks greater than
15 mm (i.e. at least 15 mm in depth) during service. This leads to the argument that the
critical crack size should be at least 15 mm. Moreover, for a given initial crack length,
considering a 5 year service time, the initial crack must survive the cyclic service loads (i.e. 5 x
7500 = 37500 cycles). If the initial crack becomes a repairable crack (at least 15 mm) and the
pipe can tolerate such a crack, then you can assume that the service life of the pipe can be
extended to 50 years with the maintenance and repair plan proposed. So, in short, you need
to consider a crack size of at least 15 mm and check whether the pipe can tolerate the water
hammer condition (if it occurs in that 5 year maintenance and repair interval). This means
that you need to apply the failure assessment diagram and check whether the pipe will be
safe or not for a specific wall thickness with a repairable crack size.

Let us think about question (e). Again, as argued before, you need to identify critical
conditions when designing safety critical structures. You already know that residual stresses
(which are self-equilibrating stresses when there are no external loading is present) can be
tensile or compressive in nature (both tensile and compressive stresses will be present near

Page 2 of 10
MACE43001/MACE 61057: Structural Integrity Lecturers: Dr. KB Katnam & Dr. AP Jivkov

welds as self-equilibrating stresses). Now, if you have some internal pressure in the pipe, you
have to add/subtract residual stresses to/from the applied stresses at every material point.
This means that you need to consider residual stresses (especially tensile residual stresses as
they add to the applied tensile stresses and the combination represents a critical condition
at some material points in the pipe).

5. FAILURE ANALYSIS AND DESIGN

The basic idea is to choose either material A or B and provide a wall thickness for ensuring
the design requirements together with the maintenance and repair plan given.

5.1 Allowable Stresses

This is the stress level that is allowable at every material point for a given material type. It is
defined in terms of the material yield strength (𝜎𝑌 which is different for material A and B).
Defining a dimensionless coeffieciant 𝑓 such that 0 < 𝑓 < 1, then the allowable stress will
be 𝑓𝜎𝑌 . As a design requirement, the allowable stress must be ≤ 0.6𝜎𝑌 (i.e. 𝑓=0.6).

5.2 Hoop Stresses

If the wall thickness, the internal diameter, and the internal pressure are 𝑡, 𝐷 & 𝑝 (note
𝑝 = 𝜌𝑔ℎ), respectively, then the hoop stress (𝜎𝐻 ) in a thin-walled pipe (this is our
assumption) is,

𝑝𝐷
𝜎𝐻 =
2𝑡
(1)

Figure 1

5.3 Stress Intensity Factor

For a given remotely applied stresses (𝜎) (i.e. here the hoop stress, 𝜎𝐻 ) and a crack size (𝑎𝑖 ),
the mode-I stress intensity factor is defined as 𝐾𝐼 = 𝑌𝜎𝐻 √𝜋𝑎𝑖 . For the pipe with an edge
crack (see Fig. 1), the geometric correction factor (𝑌) need to be evaluated as a function of
the ratio of the crack depth and the wall thickness, i.e. 𝑎/𝑡. To avoid catastrophic fracture,
the stress intensity factor (𝐾𝐼 ), when considering the combination of tensile residual
stresses and hoop stresses, must be less than the critical stress intensity factor or material
fracture toughness (𝐾𝐼𝑐 ), i.e. 𝐾𝐼 < 𝐾𝐼𝑐 .

The geometric correction factor (𝑌) for a plate with an edge crack (where 𝛼 = (𝑎⁄𝑡)) is,

Page 3 of 10
MACE43001/MACE 61057: Structural Integrity Lecturers: Dr. KB Katnam & Dr. AP Jivkov

𝑌 = 1.12 − 0.23𝛼 + 10.55𝛼 2 − 21.72𝛼 3 + 30.39𝛼 4

(2)

5.4 Critical Fatigue Crack Length

You know that any given initial crack size will grow under cyclic loading conditions. In this
case, the initial crack depth will grow under cyclic hoop stresses and then reach a stage
where the crack becomes unstable for the water hammer condition (as this is a critical
condition, you need to assume that it can occur when designing against fatigue fracture).
Following the assumptions of linear elastic fracture mechanics, you need to estimate a crack
depth that may cause catastrophic fracture due to cyclic hoop stresses (again this assumes
linear elastic fracture; note that this may not be the case for the material that you choose
for the pipe; thus represents ‘a worst case’ if you choose a brittle material). The critical
fatigue crack length can be calculated by considering the case when 𝐾𝐼 = 𝐾𝐼𝑐 . Considering
the tensile residual stresses (𝜎𝑅 ) and hoop stresses together (𝜎𝐻 ),

𝐾𝐼 = 𝑌(𝜎𝑅 + 𝜎𝐻 )√𝜋𝑎𝑐 = 𝐾𝐼𝑐

2
1 𝐾𝐼𝑐
⇒ 𝑎𝑐 = ( )
𝜋 𝑌(𝜎𝑅 + 𝜎𝐻 )
(3)

If the residual stresses are ignored or negligible compared with the hoop stresses, then the
critical crack length will be,

1 𝐾𝐼𝑐 2
⇒ 𝑎𝑐 = ( )
𝜋 𝑌𝜎𝐻
(4)

For your wall thickness and material type, if the critical crack length is less than a repairable
crack size (i.e. 15 mm), then the design is not suitable for the given maintenance and repair
plan (i.e. every 5 years, and repairs for > 15 mm cracks).

5.5 Critical Fatigue Crack Growth Life:

By considering the initial crack depth and the critical fatigue crack depth, you can estimate
the crack growth or crack propagation life (i.e. the number of cycles to reach the critical
condition for fatigue fracture). If the initial crack depth (𝑎𝑖 ), the critical crack depth (𝑎𝑐 ), the
Paris’ constants (i.e. 𝑚, 𝐶), and the range of cyclic hoop stresses (∆𝜎𝐻 ) are known, then you
know from the Paris law that,
𝑑𝑎
= 𝐶(∆𝐾)𝑚
𝑑𝑁
Moreover, the range of stress intensity factor (∆𝐾) at any given time depends of the crack
depth at that time and the range of hoop stresses (note that it is given as a constant
amplitude with the stress ratio 𝑅 = 0 for this design task). This can be integrated (either
using a numerical or analytical approach) to estimate the crack propagation life for the
critical fatigue crack depth. If an average geometric correction factor (𝑌𝑎 ) is assumed (i.e.

Page 4 of 10
MACE43001/MACE 61057: Structural Integrity Lecturers: Dr. KB Katnam & Dr. AP Jivkov

the average of the geometric correction factors 𝑌𝑖 for 𝑎𝑖 ⁄𝑡 and 𝑌𝑐 for 𝑎𝑐 ⁄𝑡), then the closed
form solution for the crack propagation life is,
𝑚 𝑚
(1− ) (1− )
𝑎𝑐 2 − 𝑎𝑖 2
𝑁𝑝 = 𝑚 𝑚
𝐶 (1 − 2 ) (𝑌𝑎 ∆𝜎𝐻 √𝜋)
(5)

If this estimated fatigue propagation life represents the number of service cycles for a
period of 5 years or more (i.e. 𝑁𝑝 ≥ 37500 cycles), then the initial crack depth will not
become critical in less than 5 years. If the estimated fatigue life (for your wall thickness,
initial crack depth and material type) is less than 37500 cycles, then the design is not suitable
for the maintenance and repair plan (i.e. every 5 years).

5.6 Failure Assessment Points:

The failure assessment diagram uses the linear-elastic fracture (driven by K), the elastic-
plastic fracture (driven by J) and the plastic collapse (driven by PL) modes in a compact
representation. For the failure assessment, you need to estimate the coordinates of the
assessment point (𝐿𝑟 , 𝐾𝑟 ) for a given set of material, geometric and loading conditions. Now,
for the given design task, if a wall thickness is chosen (based on the water hammer and
material allowable stress), then the stress intensity factor (based on the crack depth, the
corresponding geometric correction factor, and the hoop stress) can be estimated, and
further the ratio of the stress intensity factor to the fracture toughness (i.e. 𝐾𝑟 ) can be
estimated. Similarly, for the chosen wall thickness, the crack depth, and material allowable
stresses, you can estimate the ratio of the applied load to the plastic collapse load (i.e. 𝐿𝑟 ).
The assessment point (𝐿𝑟 , 𝐾𝑟 ) is thus a function of ‘the applied stress or allowable stress’,
‘the crack depth’, ‘the wall thickness’, ‘the fracture toughness’, ‘the yield strength’ and also
‘the residual stress’ (as tensile residual stresses also contribute to the stress intensity
factor). For example, if a wall thickness is chosen for a specific material type and an
allowable stress level, under cyclic loading conditions, as the initial crack depth grows with
time, the assessment point will ‘move’ with time as it is a function crack depth. The task is to
keep the assessment point inside the FAD envelope to ensure safety. In your case (see Fig. 2),
the assessment point must be inside the FAD envelope for a period of at least 5 years
(assuming every crack > 15 mm size will be repaired) to ensure that the design offers a
service life of 50 years with the given maintenance and repair plan (i.e. every 5 years, and
repairs for > 15 mm cracks).

Figure 2

For locating the assessment point, you need to estimate the values for 𝐾𝑟 and 𝐿𝑟 as follows,

Page 5 of 10
MACE43001/MACE 61057: Structural Integrity Lecturers: Dr. KB Katnam & Dr. AP Jivkov

𝐾𝐼 𝑌(𝜎𝐻 + 𝜎𝑅 )√𝜋𝑎
𝐾𝑟 = =
𝐾𝐼𝑐 𝐾𝐼𝑐
(6)

Here, this equation (6) includes the tensile residual stresses (𝜎𝑅 ), the hoop stress (𝜎𝐻 ), the
crack depth, the geometric correction factor and the fracture toughness. For example, if the
initial crack depth propagates with cyclic hoop stresses, the coordinates of the assessment
point vary with time too.

Further, applying the concept of ‘plastic collapse’,

𝑃 𝜎𝐻 𝑡 𝜎𝐻
𝐿𝑟 = = =
𝑃𝐿 𝜎𝑌 (𝑡 − 𝑎) 𝜎𝑌 (1 − 𝑎/𝑡)
(7)

Here, this equation (7) includes the hoop stresses (𝜎𝐻 ), the crack depth, the wall thickness
and the material yield strength. Note that this does not include the residual stresses. As the
stresses go beyond the material yield strength, the local pre-strain distributions which cause
residual stresses will be redistributed because of the local and global plastic strains.

6. FAILURE ASSESSMENT

As mentioned in Sections 4 & 5, the task is to come up with a suitable wall thickness which
ensures the design service life considering ‘the initial crack depth’ (𝑎𝑖 ), ‘the material yield
strength’ (𝜎𝑌 ), ‘the material fracture toughness’ (𝐾𝐼𝑐 ), ‘fatigue crack grow’ (𝑑𝑎⁄𝑑𝑁), ‘the
repairable crack size’ (𝑎𝑟 ), and ‘repair intervals’ (𝑁𝑟 ). In this case, the repairable crack size is
>15 mm and the repair interval is 5 years. The other parameters depend on your individual
parameters and the material chosen (either A or B).

To explain the design approach discussed in Section 5, let us consider individual parameters
(only 10 student sets are considered here, see Table 1), do failure assessment, and then
propose suitable individual wall thicknesses for the two material types (both A and B).

Table 1: Individual parameters

Set 𝑎𝑖 D Water head Water hammer
(mm) (m) (m) (m)
1 4.6 2.2 561 740
2 3.4 2.6 566 741
3 4.5 2.4 516 723
4 3.9 2.1 563 715
5 2.9 2.1 517 733
6 3.2 2.6 532 714
7 5.6 2.6 533 708
8 3.0 2.2 545 720
9 4.4 2.4 517 709
10 3.2 2.2 532 738

Page 6 of 10
MACE43001/MACE 61057: Structural Integrity Lecturers: Dr. KB Katnam & Dr. AP Jivkov

For the set of individual parameters given in Table 1, if you consider the initial crack depth
(which means the failure is not assessed for the repairable crack size) for failure assessment
with the two material types, the assessment points can be obtained following the
procedures discussed in Sections 5.1-5.6. The assessment points for material A and material B
with an allowable stress level of 0.6𝜎𝑌 and the tensile residual stresses 𝜎𝑅 =70 MPa are shown
in the failure assessment diagram (see Fig. 3).

Figure 3: The failure assessment diagram for both material A and material B by
considering the initial crack lengths and the allowable stress as 0.6𝜎𝑌

Table 2: The wall thicknesses and the associated critical crack lengths and fatigue
propagation cycles for the assessment points shown in Fig.3

Set 𝑎𝑖 𝑎𝑐 for A 𝑎𝑐 for B Np for A Np for B 𝑡𝐴 for A 𝑡𝐵 for B

(mm) (mm) (mm) (cycles) (cycles) (mm) (mm)
1 4.6 8.2 22.8 1850 12250 19.0 38.0
2 3.4 11.2 26.9 3015 16050 22.5 45.0
3 4.5 8.8 24.3 2560 14900 20.3 40.5
4 3.9 8.8 21.0 1675 11650 17.6 35.1
5 2.9 10.9 21.5 2005 19500 17.9 35.9
6 3.2 11.3 26.0 2975 16650 21.7 43.4
7 5.6 7.6 25.8 1375 10250 21.5 43.0
8 3.0 10.8 22.2 1840 14550 18.6 37.0
9 4.4 8.9 23.8 2390 12250 19.8 39.5
10 3.2 10.6 22.7 2345 15000 18.9 37.9

The corresponding individual wall thicknesses and the associated critical crack lengths and
fatigue propagation lives for both material A and B are given in Table 2. The columns in ‘red’
[the critical crack lengths for material A (see that all the crack lengths are less than 15 mm)
and the fatigue propagation lives for both material A and B (see that all the fatigue lives are

Page 7 of 10
MACE43001/MACE 61057: Structural Integrity Lecturers: Dr. KB Katnam & Dr. AP Jivkov

less than 37500 cycles)] indicate that the design does not meet the requirements and thus
the maintenance and repair plan is not effective to provide a service life of 50 years.

As the design parameters shown in Table 2 are not suitable for a service of 50 years (for both
material A and B), you need to reduce the allowable stresses and then repeat the analysis to
make sure both the assessment points representing both the ‘static fracture’ and the
‘fatigue fracture’ are inside the FAD envelope. In this regard, for material A the allowable
stress is defined as 0.20𝜎𝑌 and for material B as 0.40𝜎𝑌 , and the results are show in Figs. 4 &
5 and Table 3.

Figure 4: The FAD diagram for both material A and material B by considering the
allowable stress as 0.2𝜎𝑌 and 0.4𝜎𝑌 together with the initial crack lengths

Figure 5: The failure assessment diagram for both material A and material B by
considering the allowable stress as 0.2𝜎𝑌 and 0.4𝜎𝑌 together with a repairable crack
length (i.e. 15 mm)

Page 8 of 10
MACE43001/MACE 61057: Structural Integrity Lecturers: Dr. KB Katnam & Dr. AP Jivkov

The assessment points for the initial crack lengths (ai) and the allowable stresses (i.e. 0.20𝜎𝑌
for material A and 0.40𝜎𝑌 for material B) are shown in Fig. 4. As the points are inside the
envelope no failure is expected for the initial crack lengths. Moreover, the assessment points
for ‘a repairable crack length’ (i.e. 15 mm) and the allowable stresses (i.e. 0.20𝜎𝑌 for material
A and 0.40𝜎𝑌 for material B) are shown in Fig. 5. You can see that the points are still inside
the envelope and thus no failure is expected if the initial crack length grows and becomes a
repairable crack length (here 15 mm).

The corresponding individual wall thicknesses and the associated critical crack lengths and
fatigue propagation lives for both material A and B are given in Table 3. The columns in
‘green’ [the critical crack lengths for material A (see that all the crack lengths are greater
than 15 mm) and the fatigue propagation lives for both material A and B (see that all the
fatigue lives are greater than 37500 cycles)] indicate that the design does meet the
requirements and thus the maintenance and repair plan is effective to provide a service life
of 50 years.

Table 3: The wall thicknesses and the associated critical crack lengths and fatigue
propagation cycles for the assessment points shown in Fig.3

Set 𝑎𝑖 𝑎𝑐 for A 𝑎𝑐 for B Np for A Np for B 𝑡𝐴 for A 𝑡𝐵 for B

(mm) (mm) (mm) (cycles) (cycles) (mm) (mm)
1 4.6 34.2 34.2 42900 43800 57.0 57.0
2 3.4 40.5 40.5 54700 55000 67.5 67.5
3 4.5 36.4 36.4 52750 53400 60.8 60.8
4 3.9 31.6 31.6 42500 42500 52.6 52.6
5 2.9 32.4 32.4 71000 69500 53.9 53.9
6 3.2 39.0 39.0 60250 58900 65.1 65.1
7 5.6 38.7 38.7 39450 39200 64.5 64.5
8 3.0 33.3 33.3 57700 54100 55.4 55.4
9 4.4 35.8 35.8 50200 47250 59.6 59.6
10 3.2 34.2 34.2 63700 57900 56.8 56.8

7. CONCLUSIONS

Based on the analysis and results presented in Sections 5 & 6, it can be seen that the
penstock can be designed with both material A and B. However, when compared to material
A (with relatively high yield strength and low fracture toughness), material B (with relatively
lower yield strength and higher fracture toughness) provides a ‘damage tolerant’ penstock
design and thus offer an opportunity to have ‘longer inspection and repair intervals’. It is
shown in Table 3 and Figs. 4 & 5 that the allowable stresses for material B (i.e. 0.4𝜎𝑌 ) can be
much higher when compared to that of material A (i.e. 0.2𝜎𝑌 ) if a repairable crack length of
15 mm is considered with an inspection and repair interval of 5 years, ensuring a service life
of 50 years. When a higher allowable stress (i.e. 0.6𝜎𝑌 ) is considered for both material A and
B, the critical lengths are not suitable (as shown in Table 2) for the given maintenance and
repair plan.

Page 9 of 10
 MACE43001/MACE 61057: Structural Integrity Lecturers: Dr. KB Katnam & Dr. AP Jivkov

FEEDBACK: SOME COMMON MISTAKES

(a) Using the initial crack length for failure assessment without incorporating the weld
residual stresses;
(b) Using the initial crack length and residual stresses for failure assessment but without any
assessment for a repairable crack length;
(c) Estimating the fatigue crack propagation life without considering the repair intervals
(i.e. 5 years) and repairable crack length (i.e. > 15 mm);
(d) Not clearly explaining and justifying the assumptions made in the analysis;
(e) Not showing critical thinking around the concepts employed;
(f) Not reading the coursework document carefully while paying much attention to
‘additional information and hints’;
(g) Paying too much attention to ‘minor details’ and missing ‘major conceptual steps’;
(h) Poorly written reports with ‘incomplete information’;

MARKING SCHEME:

50+ 60+ 65+ 70+ 75+ 80+

Applying FAD for initial cracks 
for material A & B 
Selecting and justifying material 
B for design 
Estimating fatigue crack growth 
life for material B 
Revising wall thickness after  
fatigue life estimate 
Justifying crack shape and its   
role on Kr and Lr  
Including tensile residual   
stresses for Kr  
Applying FAD for a repairable    
crack for material B 
Revising fatigue crack growth     
life for material B 
Justifying the design for a 50     
year service life 
Justifying the maintenance and     
repair plan 
Excellent report quality and     
critical thinking 

Individual marks and feedback will be available on Blackboard (in Week 12)

Page 10 of 10