Slide 1 /140

Nodal Analysis Applied to ESP’s
Pump Sizing Exercise
Slide 1 /140
Sizing Pumps with Nodal Analysis
Say we are given the following:
Static Reservoir Pressure = 3550 psi

Flowing Wellbore Pressure = 2235 psi @ 3570 bpd
Perforations are at 7800 feet (vertical well)
Casing size is 9 5/8" and Tubing will be 3 1/2"
Wellhead pressure will be 350 psi
Electrical Power will be 50 Hz
Fluid will be 75% water with a gravity of 1.08 and
25% oil with a gravity of 28 API
There is no gas to be considered
We want to size a pump to produce 8500 bpd
Slide 2 /140
What do we do now?
Slide 3 /140
What do we do now?
It might be a good idea to see if the well is even

capable of producing 8500 bpd.
Slide 4 /140
Since we are not concerned with gas, we can use the
PI equation we derived from Darcy's law.
In this case, Pr = 3550 psi, Pwf = 2235 psi, and

Qo = 3570 bpd.
qo
PI =
( P r − P wf )
Slide 5 /140
Substituting those values into our equation gives a
PI of 2.715. With this PI we can now find out what
the maximum possible flow rate from this well will
be. Remember that Qmax will occur when Pwf is
drawn down to zero.
3570
PI = = 2.715
3550 - 2235
Slide 6 /140
We can rearrange our PI equation to solve for Qmax
Note that Qmax is greater than our target rate of 8500

bpd so the well is capable of producing the desired
flow.
Q max = PI x (P r - 0) = 2.715 x (3550 - 0)

or
Q max = 9638 bpd
Now let's find out what the flowing pressure, Pwf, will
be at the 8500 bpd flow rate.
Slide 7 /140
To do this we arrange the equation as follows:
− Q
Pwf = r
P 0
PI
Slide 8 /140
Substituting in the appropriate numbers gives us a
flowing wellbore pressure of 419 psi at our desired
flow rate of 8500 bpd.
8500
Pwf = 3550 − = 419 psi
2.715
Slide 9 /140
Although we were interested in the flowing pressure
for a flow rate of 8500 bpd, we can use this equation
to calculate the flowing pressure for any flow rate
less than the Qmax of 9638 bpd. Just for the fun of it,
let's build a table of values for flow and pressure.
Pwf = 3550 − Q 0 = ? psi

2.715
Slide 10 /140
Use the equation below Flow-BPD Pwf - PSI

to fill out the table on 0
the right. 1000
2000
3000
4000
5000
6000
Q0 7000
Pwf = 3550 −
2.715 8000
8500
9638
Slide 11 /140
Using this equation, Flow-BPD Pwf - PSI

you should come up 0 3550
with answers close to 1000 3181
those shown. 2000 2813
3000 2445
4000 2077
5000 1708
6000 1340
Q0 7000 971
Pwf = 3550 −
2.715 8000 603
8500 419
9638 0
Slide 12 /140
Remember that the values shown for Pwf indicate the
pressure of the wellbore at the depth of the
perforations. In most cases, the pump will not be set
that deep but will rather be set higher in the hole.
Since we are not concerned about gas in this example,

there is no benefit from setting the pump deeper. In
fact, all we do with increased depth is add more friction.
As a safety factor, let's set the pump at a depth such

that the intake pressure is 200 psi. This should be
enough to keep the well from pumping off.
Slide 13 /140
How do we do this? Well we know that the flowing
pressure will be 419 psi at our target of 8500 bpd.
In order to get an intake pressure of 200 psi, we will
need to raise the pump the equivalent of 219 psi.
In order to do this, we will need to know what the

specific gravity of the fluid is.
Slide 14 /140
Remember that we are producing 75% water with a
gravity of 1.08 and 25% oil with a gravity of 28 API.
First convert the API to specific gravity.
Slide 15 /140
The specific gravity for the oil is 0.8871 according to the
equation:
141.5
γo = = 0.8871
131.5 + API
Slide 16 /140
But remember that the fluid is only 25% oil. We need
to calculate the specific gravity for the fluid mixture.
We can do that with the following equation:
γ f = ( γ w × f w ) + (γ o × fo )
Slide 17 /140
The specific gravity for this particular fluid is 1.032
γ f =(1.08 x 0.75)+(0.8871 x 0.25) = 1.032
Slide 18 /140
Now that we know the specific gravity, we can convert
the pressure to depth in feet with the following
equation.
But what does this mean?
psi x 2.31 219 x 2.31

Feet = = = 490 feet
γf 1.032
Slide 19 /140
This means that we can set the pump 490 feet above
the perforations and still maintain a pump intake
pressure of 200 psi.
So what will be the pump setting depth?
Slide 20 /140
This means that we can set the pump 490 feet above
the perforations and still maintain a pump intake
pressure of 200 psi.
So what will be the pump setting depth?
The pump will be set at 7800 - 490 or 7310 feet.
Slide 21 /140
Since we are not having to worry about gas, we can
make the assumption that the fluid specific gravity
will remain relatively constant. At a flow rate of 8500
bpd, setting the pump at 7310 feet lowered the pump
intake pressure from the Pwf of 419 psi to 200 psi.
Since we are assuming the gravity will not vary

much, we can assume that the pump intake pressure
will be 219 psi lower than Pwf at all flow rates.
Slide 22 /140
By the way, we have only corrected for the weight of
the fluid column. Have we left anything out?
Slide 23 /140
YES! We have not included friction loss for the fluid
traveling up the casing to the setting depth. Should
we go back and do this?
Slide 24 /140
Usually not. We do not have a friction loss chart for
9 5/8" casing but if you look at the tubing loss chart
and extrapolate a little, it looks like the loss would be
close to zero. It is a safe bet we can ignore it in this
case.
Slide 25 /140
Let's update our chart. Flow-BPD Pwf - PSI Pint - PSI

0 3550 3331
Pint is Pump Intake 1000 3181 2962
pressure or, more 2000 2813 2594
generally, the wellbore 3000 2445 2226
4000 2077 1858
flowing pressure at a
5000 1708 1489
depth of 7310 feet.
6000 1340 1121
7000 971 752
8000 603 384
8500 419 200
Pint = Pwf - 219 9638 0 -219
Slide 26 /140 (1) This is a negative pressure

Note that at 9638 bpd, the pump intake pressure is -
219 psi. How is this possible?
Slide 27 /140
Note that at 9638 bpd, the pump intake pressure is
-219 psi. How is this possible?
It's not.
Slide 28 /140
In that case, what does this mean?
Slide 29 /140
In that case, what does this mean?
It means that this well is not really capable of

delivering 9638 bpd up to our setting depth. We
already know that the well can deliver our required
rate of 8500 bpd but what is the maximum rate the
well can deliver to the pump setting?
Slide 30 /140
The lowest pressure we can possibly achieve is 0
psi. In this case we want to know the flow rate for a
Pint of 0 psi. Can we plug this number into our PI
equation and get the new flow rate?
Slide 31 /140
The lowest pressure we can possibly achieve is 0
psi. In this case we want to know the flow rate for a
Pint of 0 psi. Can we plug this number into our PI
equation and get the new flow rate?
No we cannot. Our PI equation is only valid for the

reservoir to the wellbore and it is only valid that far
because we have made some simplifying
assumptions(i.e. no gas interference, no skin
damage, 100% perforation efficiency, etc.).
So what do we do?
Slide 32 /140
The solution is to calculate the Pwf which
corresponds to a Pint of 0 psi. In this case it is pretty
easy because we already know that the Pwf in this
case will be 219 psi. (Remember we are ignoring
friction in the casing).
We can also look at this graphically:
Q max = PI x (P r - 219) = 2.715 x (3550 - 219)

or
Q max = 9044 bpd
Slide 33 /140
Note that by raising the pump, we are effectively
shifting`the PI curve downward by 219 psi. What would
happen if we lowered the pump?
4000
3000
Pwf (@7800 ft)
Pwf
2000
Pint (@7310 ft)

1000
0
Slide 34 /140
0 2000 4000 6000 8000
Qo - BPD
This "adjusted" PI (or IPR curve if we used that
method) shows us the available pressure, at the level
where the pump intake would be, for any flow rate
from 0 to the maximum the well can produce based
on complete drawdown to zero psi at this depth. Note
we are calling this the level of the pump intake rather
than just "the pump intake". Although these two
mean the same thing, we do not want to be confusing
because, so far, we do not have a pump in the hole.
Looking at this in more general terms, this curve

represents the total amount of energy available at the
pump intake depth for any given flow rate.
Slide 35 /140
Let's make one correction Flow-BPD Pwf - PSI Pint - PSI

to our table to account for 0 3550 3331
the new Qmax. 1000 3181 2962
2000 2813 2594
3000 2445 2226
4000 2077 1858
5000 1708 1489
6000 1340 1121
7000 971 752
8000 603 384
8500 419 200
9044 219 0
Slide 36 /140
So far we have looked at everything from deep
inside the reservoir all the way up to the pump.
Now let's look at the problem from the opposite
direction. We know that our required wellhead
pressure is 350 psi and we will use 3 1/2" tubing.
Let's now calculate from the top down to see what
the tubing pressure will have to be (at the pump set
depth of 7310 ft) in order to lift the fluid to the
surface.
Remember when we did the TDH example we

worked in feet. In this particular case, we will
continue to work in pressure. We can easily
Slide 37 /140
convert from one to the other.

For this particular part of the problem, we want to
calculate the pressure at the bottom (or entrance) of
a piece of pipe. What factors will we need to
consider to calculate this pressure?
Slide 38 /140
If the pipe were laying horizontally and we knew the
pressure at the discharge, all we would have to do to
calculate the pressure at the entrance would be to
determine the pressure drop in the pipe. Since the
pipe does not change size and the fluid is
incompressible, the only loss we have is due to
friction.
∆Pfriction
Pdischarge
Slide 39 /140
Pentrance = Pdischarge + ∆P
∆ friction
Pdischarge
If the pipe were vertical and we knew
the pressure at the discharge, we
would have to consider the weight of
the column of fluid as well as the
friction. There is nothing else to be
∆Pfriction
considered. These are the only
∆h
factors which will affect the pressure
at the pipe (tubing) entrance.
∆Pgravity = ρg∆
∆h
Pentrance = Pdischarge + ∆P
∆ friction + ∆Pgravity
Slide 40 /140
Remember in the TDH example problem we had to
consider the fluid level in the annulus and subtract
this level from our setting depth to calculate a net
lift. Do not confuse that problem with this. The
difference here is that now we are not looking at the
entire well system -- we are only looking at a piece of
pipe and trying to calculate an unknown pressure at
some point in the pipe based on a pressure we do
know at some other point.
Slide 41 /140
We already know the discharge pressure (wellhead
pressure is 350 psi). Next let's determine the friction
loss from our chart.
Slide 42 /140
We start on the X-axis at 8500 bpd and read up to the 3 1/2" tubing line. We
then read the friction loss in feet/1000' over on the Y-axis. In this case, the
friction loss will be about 195 feet/1000'
Slide 43 /140
From the chart, we were able to determine that the
friction loss would be 195 feet/1000'. We can convert
that to pressure with our equation:
195
Feet Friction = x 7310 = 1425 ft
1000
or
1425 x 1.032
PSI = = 636 psi
2.31
Remember that the chart reads in loss per 1000 feet

of pipe. We must multiply by the total length to get
the actual loss.
Slide 44 /140
Let's make some

additions to our table Flow-BPD Pint - PSI Pwh - PSI Pfriction - PSI
0 3331 350
and add the new
1000 2962 350
information. As long 2000 2594 350
as we are at it, we can 3000 2226 350
4000 1858 350
fill out the rest of the
5000 1489 350
friction column in the 6000 1121 350
same manner. 7000 752 350
8000 384 350
8500 200 350 636
What is the friction 9044 0 350
loss at zero flow?
Slide 45 /140
Using the friction chart

and our equations, Flow-BPD Pint - PSI Pwh - PSI Pfriction - PSI
0 3331 350 0
we should get
1000 2962 350 16
answers similar to 2000 2594 350 42
those shown here. 3000 2226 350 95
4000 1858 350 163
5000 1489 350 245
636 psi is a lot of 6000 1121 350 327
friction -- What could 7000 752 350 441
we do to reduce it? 8000 384 350 587
8500 200 350 636
(Maybe Teflon coat 9044 0 350 702
the tubing?)
Slide 46 /140
Well this accounts for two of the three items we
need. We still need to determine the pressure
difference due to gravity. This is very simple as all
we have to do is calculate the pressure caused by a
column of our fluid which is 7310 feet tall. We can
do this with our pressure conversion equation since
we know the specific gravity.
Slide 47 /140
The pressure caused by the fluid column is 3265 psi.
7310 x γ f
= 3265 PSI
2.31
What this means is that, if we had a tubing string full
of our fluid and the pressure at the top (discharge)
were zero, the pressure read at the bottom would be
3265 psi. This pressure is caused by the weight of
the fluid in the string.
Slide 48 /140
Let's again update our table with this new piece of
information. Will the pressure due to gravity change
as the flow rate changes?
Slide 49 /140
The last three columns give us the 3 components of tubing intake

pressure (in psi). We can add these together to get the actual
pressure. Let's do that to simplify the table.
Flow-BPD Pint - PSI Pwh - PSI Pfriction - PSI Pgravity - PSI
0 3331 350 0 3265
1000 2962 350 16 3265
2000 2594 350 42 3265
3000 2226 350 95 3265
4000 1858 350 163 3265
5000 1489 350 245 3265
6000 1121 350 327 3265
7000 752 350 441 3265
8000 384 350 587 3265
8500 200 350 636 3265
9044 0 350 702 3265
Slide 50 /140
We now have 2 columns of
pressures. One column shows Flow-BPD Pint - PSI Pdis - PSI
what the Pump intake (annulus 0 3331 3615
pressure at the pump set depth) 1000 2962 3631
pressure would be for a given flow 2000 2594 3657
rate and the second shows what 3000 2226 3710
the pump discharge pressure 4000 1858 3778
(tubing internal pressure at pump 5000 1489 3860
set depth) would have to be to lift 6000 1121 3942
that fluid to the surface. 7000 752 4056
8000 384 4202
Remember that we have not placed
8500 200 4251
a pump in the well at this point.
9044 0 4317
We have simply calculated the
pressure at one point from two
different directions.
Slide 51 /140
The pressure at the pump intake is the amount of
energy available from the formation. The pressure at
the tubing intake (would be the pump discharge) is a
measure of the amount of energy required to lift the
fluid to the surface. Remember our rule that two
different pressures cannot exist at any one point in
space. But this is exactly what we have calculated.
For one point in space, we have calculated two
different pressures -- one from the reservoir up and
another from the wellhead down.
What does this mean?

Slide 52 /140
Let's look at the problem graphically. What this means is
that this system cannot exist. There is no flow rate at
which this system can produce fluid to the surface as it is.
4000
Prequired
P
r 3000
e
s
2000
s Pavailable
u
r 1000
e
Slide 53 /140 0
0 2000 4000 6000 8000
Qo
What would happen if these two lines actually crossed as
shown below?
4000
P
r 3000
e
s
s 2000 Prequired
Pavailable
u
r 1000
e
Slide 54 /140 0
0 2000 4000 6000 8000
Qo
This is a very real possibility. This would tell you that this
well should freely flow and, where these two lines cross is
the flow rate it will produce. In this case, the well should
produce about 3000bpd.
4000
P
r 3000
e
s
s 2000 Prequired
Pavailable
u
r 1000
e
Slide 55 /140 0
0 2000 4000 6000 8000
Qo
In our case the lines do not cross and this well will
not freely flow at any rate. But we can still get more
information from the chart. Remember that the
annulus pressure is the amount of energy we have
available to us and the tubing pressure is the
amount of energy we need. If we were somehow
able to supply some additional pressure (energy) to
this system to match the pressure requirement, we
should be able to get this system to flow.
This is exactly what a pump does.
Slide 56 /140
We can use our chart to show this pressure difference.
We can also use our table to calculate this pressure
difference.
4000
Prequired
P
r 3000
∆Prequired
e
s
2000
s Pavailable
u
r 1000
e
Slide 57 /140 0
0 2000 4000 6000 8000
Qo
We have previously calculated

the pressures from the top - Flow-BPD Pint - PSI Pdis - PSI
down and the bottom - up but 0 3331 3615
we are really interested in the 1000 2962 3631
DIFFERENCE between the two. 2000 2594 3657
We can subtract the Pint from the 3000 2226 3710
4000 1858 3778
Pdis to get what we will call Psys.
5000 1489 3860
Let's do that now.
6000 1121 3942
7000 752 4056
8000 384 4202
8500 200 4251
9044 0 4317
Slide 58 /140
We sure have gone to a lot of

Flow-BPD Psys - PSI
trouble to calculate a bunch of
0 284
points we are not really 1000 669
interested in. Psys is the "total" 2000 1063
system pressure which we 3000 1484
would have to supply in order to 4000 1920
get the well to flow at a rate from 5000 2371
0 to 9044 bpd. Let's plot this. 6000 2821
7000 3304
8000 3818
8500 4051
9044 4317
Slide 59 /140
Let's plot only our "system" pressure now. Notice that
the line has a slight upward curvature. What causes this?
Could this line ever be straight?
4000
P Straight Line
r 3000 for Reference
e
s Psys
s 2000
u
r
e 1000
Slide 60 /140 0
0 2000 4000 6000 8000
Qo
We will define this new curve as the "well system
curve" or just "system curve" for short. The reason
for this is that this curve includes everything in the
system from the reservoir to the wellhead except for
the pump. This curve is very useful to have because
now, rather than having to repeat calculations if we
want to size a different pump or a different flow, we
have already done all the work. No matter what pump
we decide to put in the well, this system curve will not
change.
This curve can also allow us to "analyze" a pump's

performance if it is already installed in the well.
Slide 61 /140
What we have just done is called "Nodal Analysis".
This sounds complicated but is really quite simple.
In Nodal analysis, we simply take a complex system
and break it down into small parts which are easier
to calculate. The wellhead was one node, the
reservoir was another node. What other nodes did
we use?
Slide 62 /140
We also used the perforations as a node and the
pump set depth (pump intake level) as a node. We
chose to break our system in two halves and we
made that break at the pump. Why did we do this?
Slide 63 /140
Because it is the pump we are interested in sizing.
We know what everything else in the well is and we
use that information to size the pump. Could we
have "broken" the system at another node?
Slide 64 /140
In nodal analysis, we can break the system apart at
any node we want to. The reason for this is the
fundamental principal that at any point in space,
only one unique pressure can exist. No matter
which direction we approach the node from, we must
arrive at the same pressure for the system to be real.
Slide 65 /140
Remember that we have already developed a system
curve for this problem which accounts for everything
but the pump. Further, this curve will not change no
matter what pump we place in the well. All we have
to know is that we want a pump which will produce
8500 BPD and we are using 50 Hz power.
In this particular case, we are going to use a

GN10000.
Slide 66 /140
This is a curve for a GN10000 pump. Note that we have

plotted the Y-axis as pressure rather than head. We could
do it either way but, since our system curve is in PSI, we
might as well be consistent.
GN10000 Single Stage Head-Capacity (60 Hz)
20
Pressure - PSI (γγ=1.032)
15
10
0
Slide 67 /140 0 4000 8000 12000
Flow Rate - BPD
Recall that the required system pressure from our table at
8500 bpd was 4051 psi. From the chart below we can see
that one stage of a GN10000 puts out about 15.15 psi (we
read that curve pretty accurately didn't we).
GN10000 Single Stage Head-Capacity (60 Hz)
20
Pressure - PSI (γγ=1.032)
15
10
0
Slide 68 /140 0 4000 8000 12000
Flow Rate - BPD
If our total need is 4051 psi and one stage only puts
out 15.15 psi, what do we do?
Slide 69 /140
Remember that, for a centrifugal pump, the head (or
pressure in this case) is additive. We can determine
the number of stages by dividing the total pressure
by the pressure per stage:
4051 psi
# of Stages = = 267 Stages
15.15 psi stage
Slide 70 /140
We have two different
curves here: one shows 4000
the total pressure P
3000 Psystem
required and the other s
shows the pressure i 2000
available from one 1000
stage.
We need to put them 0
0 2000 4000 6000 8000
on the same basis. GN10000 Single Stage Head-Capacity (60 Hz)
20
Pressure - PSI
15
10
Slide 71 /140
0
0 4000 8000 12000
Flow Rate - BPD
We can do this one of
two ways: we can either 4000
take the System curve P 3000 Psystem
and divide it by 267 to s
put it on a "per stage" i 2000
basis or we can take the 1000
Pump curve and
multiply it by 267. 0
0 2000 4000 6000 8000
Either way, we will GN10000 Single Stage Head-Capacity (60 Hz)
20
accomplish the same
Pressure - PSI
result. 15
10
Slide 72 /140
0
0 4000 8000 12000
Flow Rate - BPD
Here is a combined graph of the System curve and the
Pump curve on a 267 stage (total pressure) basis. The
System curve is the pressure required while the Pump
curve
is the pressure available. What do you think will happen
where they intersect?
267 stage GN10000 Pump
6000
5000 Ppump
Pres sure - PSI
4000
3000 em
s t
Psy
2000
1000
Slide 73 /140 0
0 2 4 6 8 10 12 14
Flow Rate - BPD (Thousands)
Where the two curves cross is exactly where this system
will operate. It will exactly produce 8500 bpd -- no more
and no less. This goes all the way back to the basic
premise that only one unique pressure can exist at any
point in space.
267 stage GN10000 Pump
6000
5000 Ppump
Pres sure - PSI
4000
3000
ys tem
Ps
2000
1000
8500 bpd
Slide 74 /140 0
0 2 4 6 8 10 12 14
When we plot this on a "single" stage basis, we get exactly
the same result. We can use whichever method is most
convenient. Plotting total pressure has the advantage that
you can read the maximum pressure directly. The single
stage basis is convenient since you do not have to change
scales when the number of stages changes.
Single stage GN10000 Pump
20
Ppump
Pres sure - PSI
15
tem
10 sys
P
5
8500 bpd
Slide 75 /140
0
0 2 4 6 8 10 12 14
Now we can begin to see the advantage of going to all the
bother to generate the System curve. We decided that we
needed a 267 stage GN10000 to produce this well. What
would happen if we only had a 200 stage GN10000
available?
6000
5000 Ppump - 267 stage

Psystem
Pressure - PSI

3000
2000
1000
8500 bpd
0
Slide 76 /140 0 2 4 6 8 10 12 14
Just by looking at the curves we can see that a 200 stage
pump will only produce about 6900 bpd. It may be
slightly more or less because we cannot read the chart
that accurately but it is close enough for government
work.
6000

Psystem
Pressure - PSI

3000
2000
1000
6900 bpd
8500 bpd
0
Slide 77 /140 0 2 4 6 8 10 12 14
Take note of how the pump head-capacity curve changed

shape. Why did it do this? Why didn't the curve just drop
down a little bit?
6000

Psystem
Pressure - PSI

3000
2000
1000
6900 bpd
8500 bpd
0
Slide 78 /140 0 2 4 6 8 10 12 14
The shift is just simple math. At wide-open flow, the
head is zero. Any number of stages times zero is
still zero. As you move toward shut-in flow, the
head/stage is a larger number so multiplying it by a
larger number of stages gives a faster increase in
total head.
Slide 79 /140
Remember that we had the option of either plotting
the curves by total pressure or on a "per stage"
basis. The curve we just saw was based on total
pressure and we already discussed some of the
advantages of this method. Let's do the same thing
again only on a "per stage" basis.
Slide 80 /140
Note that we get the exact same result in terms of where

the lines cross but notice that, instead of two Pump
curves, we have now plotted two System curves. What is
really interesting to note is how the System curve slope
changes.
25
Psystem-200 stages
20
Ppump - 1 stage
Psystem-267 stages
15
10
0
Slide 81 /140 0 2 4 6 8 10 12 14
Thousand
s
In reality, the curve is not "changing slope" but the way it
relates to the Pump curve looks like a slope change.
What you should always remember is that dividing the
System curve by a larger number of stages causes it to
"flatten out". This is always true. We will see why this is
important when we get to VSD sizings.
25
Psystem-200 stages
20 Ppump - 1 stage
Psystem-267 stages
15
10
0
0 2 4 6 8 10 12 14
Slide 82 /140 Thousands
Well we have gone to a lot of trouble to design a
pump. Generating the system curve takes more time
but provides more flexibility after the fact because,
once done, the System curve stays constant
regardless of the type of pump chosen. We can use
this curve and plot different types of pumps or
different numbers of stages on it to "analyze" how
that pump will perform.
Although the pump will not change the System

curve, is there anything that will cause it to change?
Slide 83 /140
Of course there is. Remember that the System curve
is based on reservoir, surface and fluid properties. If
these should change, it will affect the System curve.
For example, a different water cut or different
wellhead pressure will alter the System curve. Even
dropping reservoir static pressures will impact on
the System curve. In fact, another advantage of
developing these curves is that different System
curves can be plotted on one Pump curve to show
how the production from the well will change over
time.
Slide 84 /140
We will look at that a little later but, for now, based
on the one System curve we generated, we were able
to properly size a pump --- or were we?
Slide 85 /140
We made one mistake: we sized the pump based on
a 60 Hz pump curve but we were told that the power
supply would be 50 Hz. What should we do?
Slide 86 /140
We might size up a 60 Hz generator set to go with
the pump to cover the error or we might go back and
correctly size the pump. Fortunately for us we
already have the System curve so all we have to do
is adjust the Pump curve.
Slide 87 /140
There is a significant difference between a 60 Hz
curve and a 50 Hz curve. For the pump we have
chosen, we will only produce about 6150 bpd at 50
hz.
267 Stage GN10000 Pump
6000
5000 P60hz Psystem

4000
P50hz
3000
2000
1000
6150 bpd
0
Slide 88 /140 0 2 4 6 8 10 12 14
Thousands
Where does a 50 Hz curve come from anyway? Do
we just test the pump at 50 Hz and see where the
curve falls?

6000
5000 P60hz Psystem

4000
P50hz
3000
2000
1000
6150 bpd
0
Slide 89 /140 0 2 4 6 8 10 12 14
Thousands
We could do that and certainly those kinds of tests
are done but we can and do take advantage of some
general principles regarding centrifugal pumps in
order to predict how the curve will look at different
frequencies other than 60 Hz.
Before we get started into this, we should be a little

more clear. Actually the pump does not care what
the "frequency" of the power system is. Why is
this?
Slide 90 /140
This is because pumps are not electric -- they are
mechanical. The pump only cares about how fast it
is turning. In the SPS industry, we often interchange
frequency and speed since, due to the limited scope
of the typical SPS type equipment, this usually does
not cause a problem even though it is technically
incorrect.
Even though we will continue to talk in terms of

power frequency when predicting pump
performance, do not lose sight of what is actually
going on.
Slide 91 /140
The pump performance changes with rotational
speed. We are already familiar with a typical 60 hz
curve. When we talk about a 60 Hz curve, what we
really mean is a curve based on a pump rotational
speed of 3500 RPM. This is because the typical SPS
motor turns about 3500 RPM when fed 60 Hz power.
When we change the power frequency, what we are

directly changing is the motor speed. Since the
motor is attached to the pump, when the motor
changes speed, so does the pump.
Slide 92 /140
For a typical SPS motor:
Power Frequency = 60 Hz ⇒ Speed ≈ 3500 RPM
Power Frequency = 50 Hz ⇒ Speed ≈ 2917 RPM
So when we talk about a 50 Hz pump curve, we are

really talking about a pump which is turning around
2917 RPM. For a 60 Hz pump curve, we really mean
a pump which has a rotational speed of 3500 RPM.
Slide 93 /140
So what happens when a pump slows down? Are
we putting less energy into it?
Yes we are. It takes less energy to turn a pump more

slowly. If you do not believe this, how much energy
do you think it takes to not turn a pump at all?
Slide 94 /140
Since we are putting less energy into the pump, it is
only fair to expect less work out of the pump. This is
indeed what happens and is evidenced by the two
curves.
6000
5000 P60hz Psystem

4000
P50hz
3000
2000
1000
6150 bpd
0
Slide 95 /140 0 2 4 6 8 10 12 14
Thousands
The 50 Hz curve puts out both less flow and head.
But how can we predict how much less flow and
head? This is a problem that needs a solution so
that we can know how to size the pump.
6000
5000 P60hz Psystem

4000
P50hz
3000
2000
1000
6150 bpd
Slide 96 /140 0
0 2 4 6 8 10 12 14
Thousands
As it turns out, centrifugal pumps obey, over a very
wide range of rotational speeds, what are known as the
affinity laws. These laws are as follows:
RPM 2
Flow RPM 2 = Flow RPM 1 ×
RPM 1
2
RPM 2
Head RPM 2 = Head RPM 1 ×
RPM 1
3
RPM 2
BHPRPM 2 = BHPRPM 1 ×
RPM 1
These laws are correct because they are based solely

on rotational speed. We can express these laws in
terms of frequency as long as we remember that we are
Slide 97 /140
really defining speed.

These laws will be the basis for all adjustments of
pump curves for other power frequencies. From these
equations we can see that, if we know the head, flow
and BHP at 60 Hz, we can easily calculate what the
corresponding values will be at any other frequency.
HZ
Flow HZ = Flow 60 ×
60
2
HZ
Head HZ = Head 60 ×
60
3
HZ
BHP HZ = BHP60 ×
60
Slide 98 /140
This is a very important point. Before we can find the
performance at some other frequency, we first need to
know what the performance is at 60 Hz. These affinity
law equations are our road map to get where we want
to go. The 60 Hz curve is our "You Are Here" sign. If
we don't know where we are, we can't get where we are
going even with a map.
HZ
60
2
HZ
60
3
HZ
BHP HZ = BHP60 ×
Slide 99 /140
60
GN10000 Single Stag e 3500 RPM Performance Curve

Head HP Pump
Feet Motor Only
Head-Capa You Load Eff.
city ar e He
40 re
30 60
c y 5 50
en
f ici 4 40
20 Ef oad
u mp O nl y L
P 3 30
2 20
10
1 10
0
0 2000 4000 6000 8000 10000 12000 14000
Slide 100 /140
Capacity - BPD
Let's use what we know to size the number of stages
required for this GN10000 pump at 50 Hz. We know
that we want to produce 8500 BPD. Is this a "50 Hz"
flow or a "60 Hz" flow?
Slide 101 /140

8500 BPD is the flow we want to produce at 50 Hz. So,
before we can get the head from the 60 Hz curve, we
need to know what 60 Hz flow rate corresponds to our
50 Hz flow rate of 8500 BPD.
We do this with the affinity law equation for flow.

HZ
60
This will not work exactly like it is because the Flow60
is what is unknown. We can rearrange the equation:
60
Flow 60 = Flow 50 ×
50
Note we have substituted "50" in place of "HZ"

Slide 102 /140
According to this equation, the 60 Hertz flow rate
which is equivalent to 8500 BPD at 50 Hz will be
10,200 BPD. This is the flow rate we must use to read
the head from the curve.
GN10000 60 Hz Single Stage
25
20
15
Move up to the 60 Hz curve and
10 read head per stage (13.0939 psi)
5
Correct for 60 hz to get flow
Start at 8500 BPD (in this case 10,200 bpd)
Slide 103 /140
0
0 2 4 6 8 10 12 14
Thousands
The head per stage at 60 Hz at the equivalent flow is
13.0939 psi. We can now correct this head to 50 hz
with our affinity laws.
GN10000 60 Hz Single Stage

25
20
15
Move up to the 60 Hz curve and
10 read head per stage (13.0939 psi)
5
Correct for 60 hz to get flow
Start at 8500 BPD (in this case 10,200 bpd)
Slide 104 /140
0
0 2 4 6 8 10 12 14
Thousands
The 50 hz head will be 9.093 psi.
2
HZ
60
or
2
50
Head 50 = 13.0939 × = 9.093 psi
60
Slide 105 /140

We now know that the GN10000 will produce 9.093
psi per stage at a flow rate of 8500 bpd and 50 Hz.
We still need 4051 psi from the pump because the
System curve is not affected by pump size, frequency
-- anything at all. If we want this well to produce 8500
bpd, we had better come up with 4051 psi somehow.
How many stages will we need at 50 Hz?
Slide 106 /140

4051 psi
# of Stages = ---------------------- = 446 stages
9.093 psi/stage
Remember at 60 Hz, we only needed 267 stages. Now

at 50 Hz, we need almost 200 more. Why is this?
Slide 107 /140

We need more stages at 50 hz because the pump is not
capable of doing as much work when it turns more
slowly.
What would happen to the number of stages if we were

running at 70 Hz?
Slide 108 /140

You would think we would need less than 267.
You would probably be right.
Slide 109 /140

The blue dot represents the actual operating point of
our 446 stage pump at 50 Hz. We got this point by
using the 60 Hz curve and correcting for speed
(hertz). GN10000 Single Stage Curve
25
20
60 Hz Curve
Equivalent Curve Point
15
10
Operating Point at 50 Hz,
8500 BPD = 9.09 psi
5
0
Slide 110 /140 0 2 4 6 8 10 12 14
Thousands
The affinity laws apply for every point on the curve --
not just the one we calculated. Let's calculate some
other points. For starters, let's try shut-in.
GN10000 Single Stage Curve
25
Shut-in - Flow = 0, PSI = 19.36

20
60 Hz Curve
15
10
0
Slide 111 /140 0 2 4 6 8 10 12 14
Thousands
Using the affinity laws, we calculate flow as still
being zero but head is now 13.45 psi. Let's add this
point to our graph.
50
Flow 50 = 0 × =0
60
2
50
Head 50 = 19.36 × = 13.45 psi
60
Slide 112 /140

We can add as many points as we want. Let's try
4000 bpd (16.89 psi) and 13,000 bpd (7.62 psi).

25
20
60 Hz C
u rve
15
10
0
Slide 113 /140 0 2 4 6 8 10 12 14
Thousands
The 50 Hz point for 4000 bpd will be 3333 bpd and
11.73 psi. For 13,000 bpd, it will be 10,833 bpd and
5.29 psi. Let's add these points to our graph.
50
Flow 50 = 4000 × = 3333
60
2
50
Head 50 = 16.89 × = 11.73 psi
60
50
Flow 50 = 13000 × = 10833
60
2
50
Slide 114 /140
Head 50 = 7.62 × = 5.29 psi
60
We can continue to add as many points as we want.
Eventually we can get enough points to draw a line
through and have a "50 Hz curve".
25
20
60 Hz C
u rve
15
10
0
Slide 115 /140 0 2 4 6 8 10 12 14
Thousands
Here is the full 50 Hz curve. Affinity law
corrections are one area where many people get
confused so we want to spend a little more time on
them here just for practice.
25
20
60 Hz
Curve
15 50 H
z Cu
rve
10
0
Slide 116 /140
0 2 4 6 8 10 12 14
Thousands
Remember the affinity laws will work for many
frequencies -- not just 50 and 60 Hz. Let's take one
point on our 60 Hz curve and calculate how the pump
will be affected for several frequencies.
By the way, up till now we have been looking at a

GN10000 curve in terms of PSI. Most catalog curves
are published in "feet of head" (why is this?). The
affinity laws do not care what the units are -- they are
simply a convenience for us. If the pressure is
decreased by 69.4% from 60 to 50 hz, so will the feet of
head (sp.gr. does not change). Let's go back to feet of
head for now.
Slide 117 /140
Now we want to read one set of points from the
60 Hz curve and correct for several frequencies.
GN10000 Single Stage 3500 RPM Performance Curve

Head HP Pump
Feet Motor Only
Load Eff.
Head-Capa
40 city
30 60
ncy 5 50
c ie
fi 4 40
20 Ef ly Load
mp O n
Pu 3 30
2 20
10
1 10
Slide 118 /140

0
0 2000 4000 6000 8000 10000 12000 14000
Capacity - BPD
Let's look at the BEP (best efficiency point) for this
pump. We can see from the curve that this is about
10,000 bpd (this is why it is called a GN10000).
Head HP Pump
Feet Motor Only
Load Eff.
Head-Capa
40 city
Head/Stage
60
is 29.97 ft. 30
ncy 5 50
ie BHP/Stage
c
20 E ffi is 3.332 4 40
mp O nly Load
Pu 3 30
2 20
10
1 10
Slide 119 /140

0
0 2000 4000 6000 8000 10000 12000 14000
Capacity - BPD
Let's review our affinity laws. These are based on
rotational speed so they are correct. We are going to
make the assumption that the pump is turning 3500
RPM at 60 Hz.
RPM 2
Flow RPM 2 = Flow RPM 1 ×
RPM 1
2
RPM 2
Head RPM 2 = Head RPM 1 ×
RPM 1
3
RPM 2
BHP RPM 2 = BHP RPM 1 ×
RPM 1
Slide 120 /140
Although we have not discussed motors yet, we are
going to learn that the motor speed changes directly
with the frequency or:
RPM 2 HZ 2 HZ 2
= or RPM 2 = RPM 1 x
RPM 1 HZ 1 HZ 1
We can substitute this into the previous equations

and derive the following: HZ
60
2
HZ
60
3
HZ
Slide 121 /140 BHP HZ = BHP60 ×
60
Using these equations, let's calculate what the head
capacity and BHP will be for this one point at 40, 45,
50, 55, 65 and 70 Hz. Remember that we are only
looking at BEP and that, at 60 Hz, the flow was
10,000 bpd, the head was 29.97 ft/stage and the BHP
was 3.332.
HZ
60
2
HZ
60
3
HZ
Slide 122 /140 BHP HZ = BHP60 ×
60
At 40 hertz, we calculate the following:
40
Flow 40 = 10000 × = 6667
60
2
40
Head 40 = 29.97 × = 13.32
60
3
40
BHP 40 = 3.332 × = 0.987
60
At 50 hertz, we would calculate: 50
Flow 50 = 10000 × = 8333
60
2
50
Head 50 = 29.97 × = 20.82
60
3
50
Slide 123 /140 BHP 50 = 3.332 × = 1.928
60
Let's calculate the remainder of our points and build
a table. While we are at it, let's go ahead and add to
this table the pump rotational speed as well as the
efficiency. Remember that the efficiency is defined
by the following:
Flow * Head * γ
η = ---------------------
BHP * 135,771
When
Flow is in bpd
Head is in feet
BHP is in horsepower
Slide 124 /140
Let's first calculate the efficiency at 60 Hz:
10000 *29.97 * 1.032

η = ----------------------------- = 68.4%
3.332 * 135771
(We could also have read it off the curve)
What will happen to the efficiency as we change

speed?
Slide 125 /140

Well, as we increase speed (hertz), we increase both
head and flow so the efficiency should increase
right?
Slide 126 /140

But wait a minute! We are also increasing the BHP
and it is increasing by the hertz ratio cubed! Since
we are dividing by this large number the efficiency
should go down.
Let's just build the table and stop guessing.
Slide 127 /140

Using our affinity laws, we should get the following.
Notice that the efficiency has not changed at all. Why
is that?
Hertz RPM Flow Head BHP Eff. %

40 2333 6667 13.32 0.987 68.4
45 2625 7500 16.86 1.406 68.4
50 2917 8333 20.82 1.928 68.4
55 3208 9167 25.19 2.566 68.4
60 3500 10000 29.97 3.332 68.4
65 3792 10833 35.18 4.236 68.4
70 4083 11667 40.80 5.291 68.4
Slide 128 /140

We show the following for 60 and ? hertz but it will
apply to any frequency. As you can see, the
frequencies exactly cancel. (Sp.Gr. = constant)
Flow 60 x Head Flow Hzx Head

We know that η 60 = BHP x K
60
and that η Hz = BHP x K
Hz
60 Hz
2
60 Hz and
but we also know that Flow = Flow x , Head = Head x
Hz 60 Hz Hz 60 60
Hz 3
BHP = BHP x . If we substitute in the RHS of these last three equations
Hz 60 60
into the 60 Hz efficiency equation, we would get the following:
2 3
Hz Hz Hz
Flow x x Head x 60
60 60 60 60 Flow 60 x Head Flow 60 x Head
η = = 60 x = 60
Hz 3 BHP x K 3 BHP x K
Hz 60 Hz 60
Slide 129 /140 BHP60 x xK
60 60
In fact, because the efficiency did not change we can
have pretty good confidence that we did our
calculations correctly.
Since we have gone to the trouble of making a table,

let's go ahead and plot our results.
Hertz RPM Flow Head BHP Eff. %

40 2333 6667 13.32 0.987 68.4
45 2625 7500 16.86 1.406 68.4
50 2917 8333 20.82 1.928 68.4
55 3208 9167 25.19 2.566 68.4
60 3500 10000 29.97 3.332 68.4
65 3792 10833 35.18 4.236 68.4
Slide 130 /140 70 4083 11667 40.80 5.291 68.4
We have drawn a locus through our points. Note the
efficiency stays flat. Also notice how the BHP rises
more steeply than the head does. Why is this?
Head HP Pump
Feet Motor Only
Load Eff.
Head-Capa
40 city
30 60
ncy 5 50
c ie
f fi 4 40
20 E ly Load
mp O n
Pu 3 30
2 20
10
1 10
0
Slide 131 /140 0 2000 4000 6000 8000 10000 12000 14000
Capacity - BPD
This efficiency "shift" is kind of interesting and it is
something which we need to be aware of. In fact, it
may save a lot of time in designing pumps if we
consider it from the very beginning.
What do we mean by that?
Slide 132 /140

Let's say we are in Germany and we need to design a
pump for a well from which we want to produce 3300
bpd.
If we were in the USA we could simply go to the

catalog and look for a GN3300 pump since the oil
well pumps are named for their best efficiency point.
Well we would not find a GN3300 but we would find a

GN3200. Close enough. Without giving it too much
thought, we would probably figure out that this
would be the pump we would want.
Slide 133 /140
The problem is we are not in the USA. We are in a
country where the power is 50 Hz. But we can use
what we know now to guess the correct pump size
anyway. Remember we want 3300 bpd. All we have
to do is calculate the equivalent 60 Hz flow rate.
Slide 134 /140

60
Flow60 = 3300 * ------- = 3960 bpd
50
If we want the most efficient "G" series pump, we

would look for a GN3960. Well we can't find one of
those but we can find a GN4000. This is much closer
than the GN3200 would be.
We have probably picked the best pump for the

application and we have not even looked at a curve
yet.
Slide 135 /140
This is a nice shortcut to keep in mind. In fact, a
good rule of thumb when sizing for 50 hz power is
simply to take the next larger pump than the one that
sounds right.
For example, if you want to produce 1300 BPD (50

Hz) in 5.5" casing, use a DN1750 instead of a
DN1300. If you want to produce 5600 BPD (50 Hz) in
7" casing, use a GN7000 instead of a GN5600.
This will almost always work.
Slide 136 /140

We showed on a previous graph how the BEP point
moved left or right depending on speed. This graph
shows how the entire efficiency curve shifts for any
flow rate.
Eff%
GN10000 Efficiency vers u s Flow
80%
40 Hz 60 Hz 70 Hz
60%
40%
20%
0%
Slide 137 /140 0 2000 4000 6000 8000 10000 12000 14000 16000 18000
Summary
In this section, we practiced sizing a pump based on a
Productivity Index. We first made sure the well could
deliver the amount of flow we wanted (Vogel will
change the Qmax).
We then calculated the flowing well pressure at a series

of flows from zero to maximum.
We then saw that, by raising the pump in the hole, we

actually reduced the maximum potential flow rate.
Slide 138 /140

Summary
We adjusted the flowing pressure to the pump
setting depth and then we attacked the problem from
the opposite direction -- i.e. from the wellhead down.
The purpose in doing this was to develop a "System

curve" for the well. This System curve accounted for
all the energy gains and loses in the well except for
the pump. This process was called Nodal Analysis.
We saw how Nodal Analysis can make a difficult task

much easier.
Slide 139 /140

Summary
Once we had this System Curve, it was a simple task
to match it to the pump curve to either size a new
pump or predict where an existing pump would
perform.
We then looked at how a change in rotational speed

affected pump performance and learned that we
could predict that change with the use of the affinity
laws if we used the 60 Hz curve as our basis.
This will be useful when we look at Variable Speed

applications.
Slide 140 /140
INFLOW AND OUTFLOW PERFORMANCE
SURFACE PRESSURE PRODUCED FLOWRATE
INJECTION GAS
At Wellhead
If Po < Pwf, the well will flow naturally
(~10% of wells by number)
If Po ≥ Pwf, the well will require Artificial Lift
(~90%)
WELL OUTFLOW
RELATIONSHIP
Po
Required Po to produce desired rate
Pwf
Reservoir Pressure- Pr WELL
INFLOW (IPR)
WELL FACE
PRESSURE Available Pwf as function of the flowrate
Slide 141 /140

INFLOW AND OUTFLOW GRAPHICALLY
Outflow
Inflow
Slide 142 /140

Slide 1 /140

Uploaded by

Copyright:

Available Formats

Slide 1 /140

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Slide 1 /140

Uploaded by

Copyright:

Available Formats

Nodal Analysis Applied to ESP’s

Pump Sizing Exercise

Static Reservoir Pressure = 3550 psi

It might be a good idea to see if the well is even

In this case, Pr = 3550 psi, Pwf = 2235 psi, and

Note that Qmax is greater than our target rate of 8500

Q max = PI x (P r - 0) = 2.715 x (3550 - 0)

Pwf = 3550 − Q 0 = ? psi

Use the equation below Flow-BPD Pwf - PSI

Using this equation, Flow-BPD Pwf - PSI

Since we are not concerned about gas in this example,

As a safety factor, let's set the pump at a depth such

In order to do this, we will need to know what the

First convert the API to specific gravity.

γ f =(1.08 x 0.75)+(0.8871 x 0.25) = 1.032

But what does this mean?

psi x 2.31 219 x 2.31

So what will be the pump setting depth?

So what will be the pump setting depth?

The pump will be set at 7800 - 490 or 7310 feet.

Since we are assuming the gravity will not vary

Let's update our chart. Flow-BPD Pwf - PSI Pint - PSI

Slide 26 /140 (1) This is a negative pressure

It means that this well is not really capable of

No we cannot. Our PI equation is only valid for the

We can also look at this graphically:

Q max = PI x (P r - 219) = 2.715 x (3550 - 219)

Pint (@7310 ft)

Looking at this in more general terms, this curve

Let's make one correction Flow-BPD Pwf - PSI Pint - PSI

Remember when we did the TDH example we

convert from one to the other.

Remember that the chart reads in loss per 1000 feet

Let's make some

Using the friction chart

The last three columns give us the 3 components of tubing intake

What does this mean?

This is exactly what a pump does.

We have previously calculated

We sure have gone to a lot of

This curve can also allow us to "analyze" a pump's

In this particular case, we are going to use a

This is a curve for a GN10000 pump. Note that we have

5000 Ppump - 267 stage

4000 Ppump - 200 stage

5000 Ppump - 267 stage

4000 Ppump - 200 stage

Take note of how the pump head-capacity curve changed

5000 Ppump - 267 stage

4000 Ppump - 200 stage

Note that we get the exact same result in terms of where

Although the pump will not change the System

5000 P60hz Psystem

267 Stage GN10000 Pump

5000 P60hz Psystem

Before we get started into this, we should be a little

Even though we will continue to talk in terms of

When we change the power frequency, what we are

Power Frequency = 60 Hz ⇒ Speed ≈ 3500 RPM

10000 29.97 1.032