Capstone for February 2021 DEWA Course #8 (Prof.

Mäkiharju)
Staff Student Assistant (SSA) Zachary Yun (zackyun@berkeley.edu) is available for advice on-line.
Note: Bonus tasks yield extra credit, but there is no requirement to attempt to solve them.

Question 1
A hydroelectric power plant shown in Figure 1 is desired to output 30MW of power. Water flows from the
open reservoir, turning the turbine and is released to atmosphere at a velocity of 3 ms . The frictional head
m3
loss in the system is 20 m and the volumetric flowrate into the turbine is 30 s . Assuming we have turbulent
flow (α = 1) and the turbine generates power with an efficiency of 85%, what is the required height of the
reservoir, z1 , to provide the desired power?

Figure 1: Hydroelectric plant problem sketch. Figure adapted from White (1998).Note that some values
have been changed.

Question 2
Vapor in a multistage flash desalination plant can carry mist droplets with salt and other contaminants,
and a mist separator is used to capture the droplets. If a 2.3 m/s flow of water vapor at 0.12 bar absolute
pressure carries the droplets to the screen, what is the largest droplet diameter the screen might
encounter?

• Assume steam is saturated

• Assume droplet is spherical and can be treated as a rigid sphere as first approximation
• Use the density and viscosity of liquid and vapor water at the appropriate saturation temperature. One
reference for such properties can be found here: NIST - Thermophysical Properties of Fluid Systems

• Hint: start with a force balance. Assume that droplet would need to travel vertically upward towards
the screen above.

Recall that for a rigid sphere, the coefficient of drag can be estimated as:

24 Re/5 (Re/263000)−7.94 Re × 10−6

Cd = + 2.6 + 0.411 −8
+ 0.25 (1)
Re (1 + (Re/5) 1.52 (1 + (Re/263000) 1 + Re × 10−6

1
Question 3
To have a robust mass flow rate measurement with a relatively low pressure loss, a venturi flow meter may
be utilized. Let us consider the performance and likelihood of cavitation in a horizontal venturi. We shall
assume constant fluid properties, and denote the contraction ratio as β, defined as β = dthroat /dpipe .

(A) Sketch the venturi and label locations where you refer to in your equations. Then, derive from con-
servation of mass and Bernoulli equation the mass flow rate as a function of fluid density, measured
differential pressure ∆p = ppipe − pthroat and β as defined above. Show that the following is true:
s
2ρ∆p
ṁ = Athroat
1 − β4

(B) As no instrument is perfect, from the lectures we recall that above equation needs to be multiplied by
a discharge coefficient. Assume it has a constant value of Cd = 0.978. Our venturi has β = 0.5 and
Athroat = 0.005m2 . We shall assume uncertainty in density is 1%, in area of throat 0.7% and in β is
0.7% for each, all with 95% confidence. If mass flow rate of 19kg/s is to be determined within 3% with
95% confidence, what is maximum uncertainty allowable for the differential pressure measurement?
Assume that β = 0.5 and Athroat = 0.005m2 .

(a) Use the Gauss method

(b) Use the Monte Carlo method
(C) If cavitation inception is expected at σ ≤ 2, for the above flow rate what is minimum pressure allowed
at the venturi inlet to likely avoid cavitation? Assume water temperature is 20◦ C. Hint: start by
defining the cavitation number and be consistent with labeling from part a.

Question 4
Assume a 400mm ID pipeline with expected mass flow rate of 600 kg/s is supplying water to an older
desalination plant. Due to corrosion the average roughness of the pipe has increased to 4 mm, from the
original 0.05 mm. Assume pumping efficiency of 80%, electricity price of 0.09 USD/kWh. And, that you are
pumping seawater at 20◦ C.

(A) What is the financial cost per year for this single pipeline? Assume that the pipe is 250 m long.

(B) If a 5 mm thick liner with roughness of 0.20 mm is installed at cost of 600 USD/meter, how long is the
payback time for the liner installation?
Also note: the liner should also block any leaks the pipeline may have had.

2
Question 5
All other things being equal, smoother pipe, larger diameter, more gradual pipe transitions etc. will lead
to a reduced pressure drop in pipe network. And, ideally none of the components should be inflicted by
cavitation. During the course we discussed pressure loss estimation based of textbook approaches, and
calculations nowadays typically handled by commercial codes. For this assignment, you will examine a
simple pipe network and study effect of uncertainty to network performance using the Monte-Carlo method
for uncertainty propagation.
Use a textbook approach (with friction factors e.g. from Moody diagram (Figure 6.13 in White) based
on given roughness White (1998)) to simulate the performance the network should have. Note that simulated
performance may deviate from measured performance not only due to uncertainty about initial dimensions
of components, but also due to installation deviating from design (e.g. extra bends and rough welds), effects
of corrosion and fouling on roughness given age of pipeline, etc.
For a complex pipe network, commercial or freeware codes would typically be used to evaluate perfor-
mance of the network. While the solution methodologies vary, the basic conservation laws remain unchanged.
Hence, let us consider a simple example pipe network and a) write our own code to solve the flow rates, b)
use WNTR simulation library in python to resolve the same network and ensure your answers match, and
c) implement a Monte-Carlo solution assuming pipe diameter has uniform distribution between 98% and
103% of nominal diameter and roughness has an exponential probability distribution with mean being the
listed value. Rerun the Monte Carlo simulation with water properties at 10◦ C and 30◦ C (in addition to the
original design point of 20◦ C) in order to evaluate the influence of temperature on system performance.

Figure 2: Pipe network to analyze. Figure adapted from Jeppson (1974).Note that values have been changed.

3
NOTE: The WNTR simulation library in python will be used for comparison in this problem
and can be installed in your python distribution. WNTR is an open source library based on
EPANET. Please see the Appendix in the main notes document or https://wntr.readthedocs.
io/en/latest/installation.html for instructions on installing and implementing WNTR.

a) Use whichever method you are most comfortable with (e.g. Hardy-Cross, Newton-Rhapson, Linearized
approach) to setup equations and then solve the flow rates and pressures throughout the network. Start
from conservation of mass and consider only the major losses for ∆p calculations (i.e. neglect minor losses
due to fittings and instruments).
b) Assuming node [1] is at 140 kPag and that water to supply nodes ([1], [5] and [7]) was delivered from
open reservoirs at 0 kPag by pumps that are 85% efficient, how much energy per day is required to deliver
the water?
c) Use WNTR to check solution of part a. Report your results.
Hint: WNTR requires at least one reservoir, so even though we know all of the flowrates into the network,
you will need to set one as a reservoir. WNTR will solve for the flow from the reservoir into the network
and you can check to confirm that it matches the given value and mass is conserved. Also, make sure that
all of your junctions and pipes have unique names.
3
Check point: To check your result, the flowrate in pipe 2 should be approximately 0.289 ms and the corre-
sponding head loss should be approximately 6.45 m.
d) Use WNTR and estimate the effect of uncertainty in pipe conditions of pumping power needed to supply
network shown in figure 2. Use your code and Monte-Carlo method for propagating uncertainty and consider
i) uncertainty in pipe roughness, ii) uncertainty in pipe length, and iii) uncertainty in pipe diameter (see
table 1). Perform a convergence study to show that you have run enough iterations for your results to have
converged. To estimate uncertainty in roughness, use an exponential distribution with the mean matching
the nominal roughness. This distribution represents a 95% confidence that the roughness will be between 0
and 3x the nominal value. Report the uncertainties in pressure throughout the system based on your MC
simulation. What is the uncertainty in required pumping power?
e) Consider what happens if water temperature ranges from 5◦ C to 30◦ C. Modify your fluid properties in the
input file for each of these two limiting cases, then rerun your MC simulation. What effect does temperature
have on your results?
f ) Comment on the importance of considering each of the three sources of uncertainty (e.g. re-run simulation
while setting one uncertainty at a time to be zero and see what effect that has)?
g) Within each pipe, calculate viscous length scale (SSA will provide example upon request), wall shear stress
and estimate if a superhydrophobic coating with damage threshold of 50 Pa of shear and manufacturable
with rms roughness from 5 to 200 microns might result in drag reduction in each pipe segment (Requires
having roughness below 5 viscous length scales. For simplicity: assume that there would be a way to avoid
entrainment of gas from the surface into the flow and a way to supply gas to the surface at pipeline pressure
– not necessarily achievable presently).
Bonus task h) What would be the achievable reduction in pumping power requirements if the pipeline were
to be one standard size larger or have smoother surfaces owing to initial material selection or maintenance?
NOTE: While writing your report, a great way to show your work is to include any code that
you have written or spreadsheets you have built in the appendix. Please do so. Also upload
your code as a Jupyter notebook file along with your report.

Clarification on expectations for academic integrity:

You are highly encouraged to collaborate, trouble-solve together and share concepts.
However, do not share final/full codes or final text. All written work should be yours.

4
 Pipe Index Diameter [mm] Length [m] Roughness [mm]
Uncertainty: ±2% Uniform ±2% Uniform Exponential(µ) (95% ≤ 3µ)
(1) 610 580 0.381
(2) 460 610 0.381
(3) 610 610 0.381
(4) 460 425 0.381
(5) 610 460 0.381
(6) 610 360 0.381
(7) 460 610 0.381
(8) 380 1000 0.381
(9) 460 400 0.381
(10) 300 500 0.254
(11) 460 200 0.229
(12) 610 460 0.381
(13) 610 460 0.381

Table 1: Pipe properties in the network being analyzed.

References
Roland W Jeppson. Steady flow analysis of pipe networks: an instructional manual. 1974.
Frank M White. Fluid Mechanics, fourth edition. McGraw-Hill Higher Education, 1998.