Applied Mathematical Modelling 47 (2017) 587599

Contents lists available at ScienceDirect

Applied Mathematical Modelling

journal homepage: www.elsevier.com/locate/apm

Mathematical model and optimum design approach of

sinusoidal pressure wave generator for downhole drilling tool
Jiafeng Wu, Rui Zhang, Ruihe Wang
School of Petroleum Engineering, China University of Petroleum, Qingdao 266580, China

a r t i c l e i n f o a b s t r a c t

Article history: Mud pulse generators have been widely used for the real-time transmission of valuable
Received 12 January 2016 directional and formation data from downholes with depths of thousands of meters. There
Revised 17 March 2017
have been numerous studies on the design of mud pulse generators in which the pressure
Accepted 21 March 2017
waves were typically nonsinusoidal. Sinusoidal waves provide improved long-distance data
Available online 30 March 2017
transmission and signal noise suppression compared with nonsinusoidal waves. Although
Keywords: sinusoidal pressure wave generators have been studied in the published literature, the in-
Sinusoidal pressure wave generators uence of the risks of clogging on the design of the generator for producing sinusoidal
Mathematical model pressure waves has rarely been considered. To generate sinusoidal pressure waves and to
Optimum design approach reduce the risks of clogging, a mathematical model for the design of a sinusoidal pres-
CFD sure wave generator is developed in this paper. The effects of the axial and radial clear-
Downhole drilling tool ances between the rotor and stator on the design of the generator are considered in the
model. An optimum design method for the generator is provided by combining the devel-
oped model and a computational uid dynamics analysis. Finally, an experimental platform
was built and experiments at frequencies 2 Hz and 10 Hz were conducted to validate the
design result. The simulation and experimental results show that the optimized pressure
waves closely approximate sine waves. Therefore, the developed mathematical model and
optimization approach can be used to design a sinusoidal pressure wave generator.
2017 Elsevier Inc. All rights reserved.

1. Introduction

A key factor in improving the drilling eciency is the transmission of the valuable directional and formation data from
thousands of meters deep downhole to the surface during the drilling process. The methods for data transmission include
mud pulse telemetry [1], acoustic wave telemetry [2], electromagnetic wave telemetry [3], and a wired drillpipe [4], among
which mud pulse telemetry is still the leader due to its lower cost and because it can reach deeper wells than other tech-
nologies [5]. In recent years, new logging and measurement technologies have made it possible to collect more useful in-
formation from the downhole [6], which raises the demand for a higher data rate to transmit all of this information to the
surface in real time.
Pressure wave generators are used to produce pressure waves in the drilling uid and transmit downhole data. They can
generate three types of pressure signals: negative pulse, positive pulse and continuous wave. Negative pulses are generated
by momentarily reducing the pressure in the drillpipe and diverting mud from the inner drillpipe to the annulus via a
valve. Positive pulses are generated in the mud column by partially blocking the ow of mud through the drillpipe via a

Corresponding author.
E-mail address: hitwu@163.com (J. Wu).

http://dx.doi.org/10.1016/j.apm.2017.03.046
0307-904X/ 2017 Elsevier Inc. All rights reserved.
588 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599

Nomenclature

Pu , Pd pressure upstream and downstream of the SPWG (Pa)

mud density (kg/m3 )
Q discharge (m3 /s)
Cd discharge coecient (-)
A ow area (m2 )
Po , Qo , Ao , Cdo upstream pressure (Pa), discharge (m3 /s), ow area (m2 ) and discharge coecient (-) while the SPWG
is fully open
Pv , Qv , Av , Cdv upstream pressure (Pa), discharge (m3 /s), ow area (m2 ) and discharge coecient (-) while the ow
area A=Av
P, P0 calculated pressure wave (Pa), reference sine pressure wave (Pa)
t, t time (s), time interval (s)
relative valve opening ()
C mean pressure value and oscillation amplitude of the sine pressure wave (Pa)
n wave number and the number of lobes of the orice (-)
rotor speed (rad/s)
initial phase (rad)
k relative intensity (-)
T wave cycle (s)
polar angle (rad)
A1 , A2 , A3 three regions of the ow area (m2 )
r polar radius (m)
R, R0 rotor radius (m), stator radius (m)
coecient for A3 (-)
s axial clearance and radial clearance between the stator and rotor (m)
correction coecient (-)
L length of the orice curve (m)
0 , xy Initial correlation coecient (-), calculated correlation coecient (-)
v a uniform velocity (m/s)
Notes: In the nomenclature, the symbol - denotes dimensionless.

valve mounted on the drillpipe. The continuous wave generator consists of a pair of rotor and stator, each of which has
multiple lobes. The rotor is driven by a motor. When the rotor rotates, a continuous wave is generated in the opening or
partially closed space between the lobes of the stator and rotor [7]. The operating frequencies of continuous wave telemetry
can reach a frequency as high as 24 Hz, so high data transmission rates are possible [8].
Although continuous wave generators have been improved in the past ten years, their signal amplitudes are less than
those generated by positive pulse systems. Until now, the types of stator-rotor orices of the generators have mainly in-
cluded rectangularly shaped, triangularly shaped, and sector lobe orice, among others [912]. The pressure waves gener-
ated by these generators are not perfectly sinusoidal and consist of high-frequency harmonic components with high power
factors. According to the propagation characteristic of periodic mud pressure waves in long wellbores, the amplitude of these
pressure waves decreases exponentially with an increase in frequency, and the energy of the pressure signal reaching the
surface is lost in the form of harmonics [13]. Based on communication theory [14], the sinusoidal carrier can effectively
suppress noise from the mud channel, transmit multiple signals simultaneously on the same channel without aliasing, and
spread over further distances. Therefore, to increase the data transmission rate and to enhance the strength of the signals
over a longer data transmission distance, a new generator with a different stator-rotor orice needs to be designed to gen-
erate nearly sinusoidal pressure waves, and this generator is dened as a sinusoidal pressure wave generator (SPWG). The
sinusoidal pressure wave generated by the SPWG is also a continuous wave. Normally, the drilling uid contains solid parti-
cles or debris. These particles should be able to pass through the peripheral orices of the stator of the generator and then
the orices of the rotor. If the axial clearance between the stator and rotor is less than a certain value, clogging will occur,
which may lead to downhole equipment failure [9]. Therefore, we need to consider the inuence of the axial and radial
clearances between the stator and rotor on the design of the SPWG.
Recently, there have been numerous studies on the design of SPWG. Jia [15] designed a stator or rotor with curved
orices. Theoretically, these generators can generate similar sinusoidal pressure wave signals. Zhidan et al. [16] designed an
improved arc-llet-line triangular valve orice based on a general line triangular valve orice. Namuq et al. [17,18] designed
a continuous wave generator in which the rotor had a fan-shaped orice and a circular hole on each lobe. He obtained
good results in his test. However, all of these studies do not consider the effects of the axial and radial clearances. The ow
through the clearances and orices of the pressure wave generator is three dimensional. It is dicult to directly calculate
the ow areas using a mathematical model. For these uid structures in a three-dimensional ow eld, many researchers
 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599 589

Pressure wave Axial gap Flow sleeve

Motor

Mud

Stator Orifice Rotor Drive shaft

Fig. 1. Local structure of a pressure wave generator.

have combined the theoretical model with a numerical analysis method to obtain the optimum design results [1922]. These
methods can also be used for the optimum design of the SPWG.
In this paper, a mathematical model for the stator-rotor design in the SPWG for use as a downhole drilling tool is de-
veloped, and an optimum approach is presented to optimize the design by combining the mathematical model and com-
putational uid dynamics (CFD) analysis. Based on the model and optimum approach, a prototype of the SPWG was man-
ufactured, and the experimental platform was built. The simulation results were validated experimentally. The results were
highly consistent each other.
The following sections of this paper are organized as follows. The mathematical model is proposed in Section 2, and the
optimum method is given in Section 3. Simulation and experimental results are provided and discussed in Section 4. Finally,
the conclusions are drawn in Section 5.

2. Modeling

The structure of the SPWG is shown in Fig. 1. It consists of a stator with several orices through which a stream of uid
passes and a rotor that rotates in an opposite direction as the stator rotates to block the ow of the drilling uid from the
orices of the stator. The rotor has several orices that are located along the ow area. When the rotor is rotating at a
constant speed, the ow area changes periodically and sine pressure waves are generated by the SPWG.
There are solid particles with different sizes in the drilling uid. If the axial clearance between the stator and rotor is not
large enough, the rotor may become jammed. Therefore, a reasonable axial clearance between the stator and rotor needs
to be determined. In addition, the outer diameter of the stator should be slightly larger than that of the rotor to prevent
clogging. The mathematical model and optimum design method to determine the axial and radial clearances for the SPWG
will be discussed in the following sections.

2.1. Relative valve opening model for sine pressure wave

The effect of gravity in the drilling uid is neglected. The pressure difference between the upstream and downstream of
the SPWG can be expressed as follows:
Q 2
Pu Pd = , (1)
2Cd2 A2
where Pu and Pd are the pressures at the upstream and downstream of the SPWG, respectively; Q is the discharge through
the valve; is the mud density; A is the area of the valve opening; and Cd is the discharge coecient.
Setting Pu = Po , Q = Qo , A = Ao , and Cd = Cdo when the SPWG is fully open and Pu = Pv , Q = Qv , and Cd = Cdv when the area
of the valve opening A is equal to a certain value Av and when the downstream pressure Pd of the valve discharging into
the atmosphere remains a constant, we can obtain
Qv 2
Pv Pd = , (2)
2Cd2v Av 2

Qo 2
Po Pd = . (3)
2 A 2
2Cdo o

Subtracting Eq. (3) from Eq. (2) yields

Qv 2 Qo 2
Pv Po = , (4)
2Cd2v Av 2 2 A 2
2Cdo o

The drilling uid is pumped into the drillpipe using a triplex reciprocating pump that can generate an approximate
constant discharge, and its density almost does not change due to its weak compressibility, so the discharge through the
590 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599

Fig. 2. Sine pressure wave in the ideal case.

valve can be approximately described as a constant value, i.e., Qv = Qo = Q. The discharge coecient is an empirical factor
and is related to the orice shape, orice size and material of the valve. The ow area does not have a signicant effect on
the discharge coecient. To reduce the number of variables in Eq. (4), the discharge coecient is assumed to be a constant
value as the rotor turns, i.e., Cdv = Cdo = Cd . Then, Eq. (4) can be rewritten as
 
Q 2 1
P = Pv (t ) Po = 1 , (5)
2Cd2 Ao2 (t )2
where = Av /Ao is the relative valve opening and (0,1]. It is noted that Av cannot be equal to zero.
The design objective of the SPWG is to produce sine pressure wave Pv that propagates upstream by periodically changing
the ow area of the SPWG, so the form of Pv in the model is selected as a sine wave with time when the rotor speed
is constant. In Eq. (3), the density , discharge Q and maximum opening area Ao are known and Po is a constant value.
Because Pv is a sine wave, it can be seen from Eq. (5) that P is also a sinusoidal wave with time or with the relative
valve opening , and P reaches a minimum value 0 when = 1 and a maximum value when is at the minimum valve
opening. Therefore, P can be expressed as

P (t ) = C cos(nt + ) + C, (6)
where n is the wave number or the number of lobes of the orice and is typically greater than or equal to 2, C is the mean
pressure value or the oscillation amplitude, is the rotor speed and is the initial phase.
When C is small, the amplitude of the pressure wave is small and the strength of the alternating force acting on the
downhole instrument is weak. Therefore, the inuence of the sine pressure wave on the downhole instrument is very small.
However, after the generated pressure wave propagates in the mud communication channel, the pressure signal may not be
detected on the ground because of acoustic attenuation, so the oscillation amplitude C needs to be large enough to obtain a
measurable pressure. The oscillation amplitude C can be determined based on the attenuation characteristics. It is assumed
that the relative valve opening is minimum and P is maximum when t = 0; from Eq. (6), we can obtain P = Ccos( ) + C,
and thus, = 0 (see Fig. 2, in which T is the period of the sine pressure wave).
Substituting Eq. (6) into Eq. (5) yields
 
nt 1/2
(t ) = 1 + 2kcos2 , (7)
2
where k = C/Po = 2CCd2 Ao2 /( Q 2 ) is the relative intensity.
in Eq. (7) is an even function with respect to t, so the sine pressure wave can be obtained no matter which direction
the rotor is rotating, which indicates that the rotor and stator have the same orice shape.
The inuence of k on is shown in Fig. 3, which is plotted at n = 4, = 31.4 rad/s and T = 1/n. For different n values,
the trend of the inuence of k on is similar, as shown in Fig. 3. From Fig. 3, we can see that a weaker relative intensity
results in a larger minimum relative valve opening and there is a minimum ow area at the minimum relative valve opening.
The minimum ow area may be formed by the radial clearances and axial clearances between the stator and rotor. From
t = 0 to T/4, the ow area of the valve increases slowly and the curve is approximately linear. After t = T/4, the curve gradient
increases rapidly. The curve tends to a line after t = 3T/4. At t = T/2, there is a larger curvature. The valve relative opening
for the sine pressure wave is nonlinear with respect to time. A smaller k corresponds to a larger gradient of the ow area
and a smaller curvature radius of the orice curves. Fig. 4 illustrates the ow area of the SPWG from which, if the curvature
radius is small, the maximum ow area will be small, which leads to a very large pressure drop at the valve and increases
the pump power when the SPWG is fully open. Therefore, k is set to be small as possible in the design. Its magnitude
depends on the amplitude of the pressure wave and minimum pressure drop through the valve. For example, to generate a
pressure wave with an amplitude of 0.7 MPa, the pressure drop is designed to 0.1 MPa. Therefore, k = 0.7 MPa/0.1 MPa = 7.
 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599 591

Fig. 3. The inuence of the relative intensity on the valve opening.

Fig. 4. The ow area of the SPWG.

2.2. Geometrical model of valve orice

In this section, the mathematical model of the orice curve will be derived based on the relative valve opening model
for the sine pressure wave. It is assumed that the discharge coecient is a constant value.
As described above, the ow area is minimum and the pressure is maximum when t = 0. The rotation angle of the rotor
can be described as = t. The orice curve can be dened in a polar coordinate system r( ); r( ) is the polar radius of
the valve orice and denotes the distance from the axial line to the intersection point between the orice curves of the
stator and the rotor where [0, /n]. As shown in Fig. 4, the total ow area can be divided into three parts: the axial ow
area A1 formed by the orice curve, A2 formed by the radial clearances and A3 formed by the axial clearances. These three
areas can be determined as follows:
  
A1 ( ) = 0.5 R2 r ( ) d ,
2
(8)
0

where R is the outer diameter of the rotor, and

A2 ( ) = 0.5 ( R0 2 R2 ), (9)
where R0 is the diameter of the stator.
The area A3 is the ow area that is formed by the clearance between the stator and rotor. It is related to the orice
shape and axial clearance s labeled by the red dot-dash line in Fig. 4. From Figs. 2 and 4, the orice curve has a signicant
inuence on A3 when the rotor rotation angle lies in the range [ , /n]. The inuence can be approximately characterized
by the variable ( ). When = 0, the valve is fully closed and the ow area is mainly formed by the axial clearances, so
(0) = 1. When = /n, the axial clearance does not affect the whole ow area of the valve and ( /n) = 0. When (0,
/n), the valve is partially open. Therefore, ( ) can be approximately described as ( ) = 1n / , and A3 can be expressed
as follows:
 /n
A 3 ( ) = s ( ) r ( )d , (10)

where ( ) is a coecient about . s is the axial clearance between the stator and the rotor. The clearance is far less than
the size of the valve orice.
592 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599

Fig. 5. The ow chart of the optimum design procedure.

Therefore, the whole ow area can be expressed as

Av ( ) = n ( A1 + A3 ) + A2 . (11)
Substituting Av = Ao ( ) into Eq. (11) yields

f ( ) = n ( A1 + A3 ) + A2 Ao ( ) = 0. (12)
Differentiating f( ) with respect to yields
kAo sin(n )
r ( )2 + 2 sr ( ) R2 + 1. 5
= 0, (13)
[k cos(n ) + k + 1]
1
where = 1 0 r[( /n )x + ]/r ( )dx is a correction coecient that can be used to correct for the inuence of the
axial clearance s on the whole ow area A. The minimum value of [( /n )x + ] is when x = 0. The maximum
value of [( /n )x + ] is /n when x = 1. Fig. 4 shows that r( ) is a strictly monotonically decreasing function, so
0 < r[( /n )x + ] /r( ) < 1 and 0 < < 1.
The solution of the equation can be expressed as

kAo sin(n )
r ( ) = s + ( s )2 + R2 1. 5
. (14)
[k cos(n ) + k + 1]

The length of the orice curve can be obtained as

 /n
L= r ( )d . (15)
0

When the valve is fully closed at t = 0, the ow area is minimum and the pressure is maximum. According to Eq. (7),
the minimum valve opening can be described as

(0 ) = (1 + 2k )1/2 = (Ls + A2 )/Ao. (16)

3. Optimum design method

Eqs. (13) and (16) provide a theoretical model for the orice curve but do not cover the effect of the three-dimensional
turbulent ow caused by using the correction coecient as a design variable. To design the SPWG and nd the effective
correction coecient, an optimum design method of the SPWG is given. The design procedure is illustrated in Fig. 5. First,
an estimated value of the correction coecient is given and the orice curve of the valve is calculated using Matlab soft-
ware based on Eqs. (14) and (16). Second, the curve is imported into Solidworks software and the 3D geometric model of
the SPWG is built. Third, the 3D model is imported into Ansys software where the nite element model is created and a
 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599 593

Table 1.
Design parameters.

Rotor Stator
Discharge Q Density Coecient Amplitude Lobes Relative radius R radius R0 Clearance s
(m3 /s) (kg/m3 ) Cd C (Pa) number n intensity k (m) (m) (m)

0.03 997 0.9 0.7 106 4 6 0.041 0.043 0.002

Fig. 6. 3-D model and mesh for CFD analysis.

transient pressure analysis is performed. Finally, the calculated results are compared with a reference sine wave using a cor-
relation coecient. The comparison is used to determine whether the analysis reaches the design goal. If not, the coecient
is changed and all of the steps are repeated. The correlation coecient can be dened as

N
[ P (i t ) P0 (i t )]
i=0
xy =  , (17)

N 
N
P2 (i t ) P0 2 (i t )
i=0 i=0

where P0 (t) is the reference sine wave, which is determined by C and Po in Eqs. (5) and (6); t is the unit time; and N t
is the length of the signal.

4. Simulation and experimental results

4.1. Simulation analysis and optimum result

The design parameters are listed in Table 1. The signal frequency was set to 20 Hz. The ow path model and nite
element meshes are provided in Fig. 6. The model consisted of the rotor, stator and a pipe. The uid model consists of ve
parts, the upstream ow, stator orice, rotor orice, axial clearance and downstream ow. The uid properties in the analysis
were set as the properties of pure water at normal temperature. The velocity inlet boundary conditions and pressure outlet
boundary conditions are selected for the simulation. Based on the ow rate and stator radius given in Table 1, a uniform
velocity v = Q/( R0 2 ) of 5.16 m/s was specied at the inlet boundary, and a zero pressure was set at the outlet boundary.
Numerical simulation was performed using the CFX module of Ansys Workbench 16.1 software, and the RNG k- turbu-
lent models were used to solve the transient ow. The pressurevelocity coupling was obtained using the SIMPLE algorithm,
and the second order backward Euler time discretization options in the CFX module were used for these simulations. The
total time was set to 0.1 s, and the time step was set to 0.0 0 078125 s, with 30 iterations per time step. The default conver-
gence criteria in the CFX module were used, root-mean-square (RMS) was selected for the residual type, residual target was
set to 104 , and value was normalized to the residual errors.
A mesh sensitivity study was performed to verify the inuence of the mesh resolution on the results. In the sensitivity
analysis, the valve orice was designed at = 0.90. Six different meshes sizes were used, as shown in Table 2. The simulated
pressure waves are presented in Fig. 7. Some observations can be made from Fig. 7. (1) The minimum values of the pressure
waves were almost the same for different mesh sizes. (2) The maximum pressures were different for the six meshes. The
deviation relative to the max pressure was less than 1.2%. Therefore, the sensitivity analysis indicates that the mesh sizes
had little effect on the numerical results. Mesh 1 in Table 2 will be used for the optimum design and analysis of the valve
orice.
In Fig. 5, a smaller step size of the correction coecient resulted in more accurate analysis results with a longer
computational time, so a reasonable step size should be selected to balance the optimization quality and computation time.
In the example given in Table 1, the step size was selected to be 0.02, and its suitableness was veried in the following
analysis results. If more accurate design results are required, a smaller step size can be selected for the correction coecient.
The analysis results are shown in Table 3. The pressure waves for = 0.96, 0.90, 0.86 and 0.8 are plotted in Fig. 8, where
594 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599

Table 2.
Six different meshes sizes for grid sensitivity analysis.

Location Upstream owpath Stator orice Rotor orice Axial clearance Downstream owpath

Mesh 1 Nodes 16,793 5229 32,799 158,275 16,793

Elements 14,309 3708 26,340 125,640 14,309
Element size (mm) 5.0 2.0 1.0 0.5 5.0
Mesh 2 Nodes 19,062 5901 37,843 230,256 19,062
Elements 17,066 4290 30,576 190,670 17,066
Element size (mm) 4.5 1.8 0.9 0.45 4.5
Mesh 3 Nodes 26,340 7704 47,905 260,820 26,340
Elements 23,836 5740 39,432 216,085 23,836
Element size (mm) 4.0 1.6 0.8 0.4 4.0
Mesh 4 Nodes 43,168 9080 60,723 403,018 43,168
Elements 39,690 6832 50,604 343,806 39,690
Element size (mm) 3.5 1.4 0.7 0.35 3.5
Mesh 5 Nodes 64,124 13,023 80,704 599,968 64,124
Elements 59,535 10,104 68,136 522,935 59,535
Element size (mm) 3.0 1.2 0.6 0.3 3.0
Mesh 6 Nodes 107,996 17,136 112,619 995,661 107,996
Elements 101,559 13,560 96,456 882,776 101,559
Element size (mm) 2.5 1.0 0.5 0.25 2.5

Fig. 7. Mesh size effect on the simulated pressure waves.

Table 3.
Correlation coecient xy at different correction coecient .

0.96 0.94 0.92 0.90 0.88 0.86 0.84 0.82 0.80

xy 0.999290 0.999300 0.999308 0.999314 0.999309 0.999273 0.999210 0.999147 0.999083

the correlation coecient achieves a maximum of 0.999314 at = 0.90. The designed orice is shown in Fig. 9, and the
prototypes of the stator and rotor are provided in Fig. 10.
Fig. 11 presents the simulation result at different frequencies based on the design result. From Fig. 11, the wave is ob-
served to be similar to a sine wave and the amplitudes of the pressure wave are almost the same for frequencies of 20 and
24 Hz. Therefore, the proposed design method is feasible according to the simulation verication. Experimental verication
will be discussed in the following section.

4.2. Experiment

The experimental equipment is illustrated in Fig. 12. The equipment primarily consists of a servomotor (SIEMENS V80),
a reducer (reduction ratio is 8:1), downhole tool assembly on which the SPWG is mounted, pipe, two pressure sensors (the
precision is 0.25%), control system for the closed loop speed control of the servomotor and NI9205 signal acquisition system,
which was used to acquire the pressure signal.
The inlet sensor and outlet sensor were used to separately measure the upstream pressure and downstream pressure of
the SPWG. After water was pumped into the inlet of the pipe, the rotor was driven by the servomotor at a uniform rotation
speed. Pressure waves at the inlet and the outlet were generated, and the pressure signal was acquired by the acquisition
system. The signal acquisition frequency was 125 Hz.
If the outlet is opened directly to the atmosphere, a ow separation or a vacuum may occur at the outlet, whereas the
uid should have a continuous ow in the pipe and the pressure at the outlet cannot be zero according to the theoretical
analysis and simulation calculations. Therefore, it is necessary to control the minimum value of the outlet pressure in the
 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599 595

Fig. 8. The analyzed pressure waves at different .

Fig. 9. The designed orice curve.

experiment, and a pressure of 1.5 MPa was applied to satisfy this condition by mounting a throttle valve at the end of the
outlet pipe. During the experiment, pressure data at frequencies of 2 and 10 Hz were collected and analyzed.

4.3. Results and discussion

Fig. 13 is the acquired signal from the pressure sensors at Q = 20 L/s and f = 2 Hz. From Fig. 13, the noise associated
with the inlet pressure and outlet pressure signals was very strong. This noise was primarily the result of pump noise. The
596 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599

Fig. 10. The prototype based on the design result.

Fig. 11. Simulation results at different frequencies.

Fig. 12. Experimental equipment.

 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599 597

Fig. 13. The original signal with a 2 Hz pressure wave.

Fig. 14. The ltered signal and amplitude-frequency characteristic of the original data at frequency of 2 Hz.

pressure difference between the inlet pressure and outlet pressure is shown at the bottom of the gure. The noise in the
pressure difference is very small because the pump noise in the inlet pressure is the same as that in the outlet pressure.
Fig. 14(a) shows the pressure differences and ltered signal at a frequency of 2 Hz in which a low-pass lter with a 3 Hz
cutoff frequency was utilized to remove high frequency noise. Fig. 14(b) shows the amplitude-frequency characteristic of
the pressure difference signal, from which we can see that the signal strength at 2 Hz is much greater than that at other
frequencies and the noise is primarily the high-order harmonics of the signal. Based on Eq. (17), the correlation coecient
was as much as 0.999508 in a single cycle.
Fig. 15 is the acquisition signal from the pressure sensors at Q = 20 L and f = 10 Hz. Similar to Fig. 13, the noise in the
sensor signal is very strong, but the noise of the pressure difference signal is small. Contrary to what is shown in Fig. 13,
the pressure difference signal in Fig. 15 is serrated and there are spiky oscillations. The data acquisition frequency for the
acquisition system was 125 Hz, frequency of the pressure signal was 10 Hz, and 12 or 13 data points were acquired in each
cycle. It is sometimes dicult to acquire the peak point of the pressure wave, and the sampling points were relatively
sparse. Therefore, the signal appears to be serrated and to have spiky peaks and other phenomena.
Fig. 16(a) shows the pressure differences and ltered signal at a frequency of 10encyin which a band-pass lter with a
[2 Hz, 12 Hz] cutoff frequency was utilized to remove high frequency noise and low frequency disturbances. Fig. 16(b) shows
the amplitudefrequency characteristic of the pressure difference signal. It is obvious that the signal strength at 10 Hz was
much greater than that at other frequencies, as shown in Fig. 16(b). Based on Eq. (17), the correlation coecient was as
much as 0.999369 in a cycle. However, the wave heights are inconsistent with each other because the signal acquisition
frequency was low and there was the same frequency interference in the original signal.
598 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599

Fig. 15. The original signal with a 10 Hz pressure wave.

Fig. 16. The ltered signal and amplitudefrequency characteristic of the original data at a frequency of 10 Hz.

The experimental results demonstrate that the designed valve can generate a sine wave at different frequencies. The
proposed mathematical model and optimum design method for the SPWG are feasible and effective.
There are differences between the experimental conditions in the paper and actual drilling conditions. For example,
the physical parameters of the drilling uid and pipe length are different. The modeling, simulation and experiments in
the paper were performed in a short pipeline (approximately 40 m), but the pipeline in the wellbore of actual drilling
is several kilometers long. The drilling uid was pure water in the paper, but was actually a liquid-particle multiphase
ow with solid particle in real application. Of the physical properties of the uid, only the density is associated with the
proposed model. From Eq. (5), the density only inuences the amplitude of the sine pressure waves and does not affect the
frequency or phase of the sine pressure waves. However, as there are multiple reection boundaries in a long wellbore, such
as the drill bit, downhole turbine, and so on, reective pressure waves may be produced and superimposed with the sine
pressure waves generated by the SPWG. When the superimposed waves propagate to the surface, their waveform features
may change due to reected waves. Therefore, the well depth may have some impact on the generation of the sine pressure
wave downhole, and this inuence needs to be further studied based on the results of this paper.

5. Conclusions

A mathematical model is presented for the SPWG to generate a sinusoidal pressure wave with a systematical consider-
ation of the inuence of the axial and radial clearances on the pressure waves in the design process to reduce the risks
 J. Wu et al. / Applied Mathematical Modelling 47 (2017) 587599 599

of clogging. A correction coecient for this inuence was embedded into the proposed model. Because the model is non-
linear and complicated to solve and there is some uncertainty in the correction coecient, an iterative approach was used
to optimize the design of the SPWG by combining the proposed model and CFD analysis. The optimization objective was
to maximize the correlation coecient for both the analyzed pressure waves and standard sine waves. The correction co-
ecient was also a design variable. An experimental platform was built, and experiments were conducted to verify the
proposed model and corresponding solution method.
The simulation results show that the pressure waves generated by the designed SPWG were all close to sinusoidal waves
at different frequencies. The amplitudes of the pressure waves were not inuenced by the signal frequency. The experimental
results show that the SPWG developed by using the proposed model and the solution method can approximately generate
sinusoidal pressure waves at 2 and 10 Hz. The amplitudes of the waves are virtually constant at lower frequencies but
inconsistent with each other at higher frequencies since the same frequency provides interference and a lower sampling
frequency relative to the signal frequency.

Acknowledgments

We would like to thank the anonymous reviewers for their valuable suggestions and corrections. This research was sup-
ported by the National Natural Science Foundation of China (Grant No. 51604296) and the Fundamental Research Funds for
the Central Universities (Grant No. 16CX02021A).

References

[1] C. Klotz, I. Wassermann, D. Hahn, Highly exible mud-pulse telemetry: a new system, in: Proceeding of the SPE Indian Oil and Gas Technical Confer-
ence and Exhibition, Society of Petroleum Engineers, 2008.
[2] S.W. Qu, K.B. Zhu, Z.P. Nie, et al., Analysis of signal transmission for use of logging while drilling, IEEE. Geosci. Remote Sens. Lett. 10 (5) (2013)
10011005.
[3] F.N. Tromenkoff, M. Segal, A. Klassen, et al., Characterization of EM downhole-to-surface communication links, IEEE Trans. Geosci. Remote Sens. 38
(6) (20 0 0) 25392548.
[4] M.A. Gutierrez-Estevez, U. Krger, K.A. Krueger, et al., Acoustic channel model for adaptive downhole communication over deep drill strings, in:
Proceeding of 2013 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), IEEE, 2013, pp. 48834887.
[5] I.N. de Almeida Jr, P.D. Antunes, F.O.C. Gonzalez, et al., A review of telemetry data transmission in unconventional petroleum environments focused
on information density and reliability, J. Softw. Eng. Appl. 8 (09) (2015) 455.
[6] C. Klotz, P.R. Bond, I. Wassermann, et al., A new mud pulse telemetry system for enhanced MWD/LWD applications, in: Proceeding of the IADC/SPE
Drilling Conference, Society of Petroleum Engineers, 2008.
[7] R. Hutin, R.W. Tennent, S.V. Kashikar, New mud pulse telemetry techniques for deepwater applications and improved real-time data capabilities, in:
Proceeding of the SPE/IADC Drilling Conference, Society of Petroleum Engineers, 2001.
[8] A. Caruzo, R. Hutin, S. Reyes, et al., Advanced design and execution techniques for delivering high data rate MWD telemetry for ultradeep wells, in:
Proceeding of the OTC Arctic Technology Conference, Offshore Technology Conference, 2012.
[9] Lavrut E., Kante A., Rellinger P., et al. Pressure Pulse Generator for Downhole Tool. U.S. Patent 6970398. 2005.
[10] Hahn D., Peters V., Rouatbi C., et al. Oscillating Shear Valve for Mud Pulse Telemetry and Associated Methods of Use. U.S. Patent 6975244. 2005.
[11] Perry C.A., Burgess D.E., Turner W.E. Rotary Pulser for Transmitting Information to the Surface from a Drill String Down Hole in a Well. U.S. Patent
7327634. 2008.
[12] Moriarty K.A. Method and System for Wellbore Communication. U.S. Patent 8020632. 2011.
[13] J. Wu, R. Wang, R. Zhang, et al., Propagation model with multi-boundary conditions for periodic mud pressure wave in long wellbore, Appl. Math.
Model. 39 (23-24) (2015) 76437656.
[14] G.P. John, Digital Communications, McGraw-Hill Companies, Columbus, 2001.
[15] P. Jia, Design of Drilling Fluid Continuous Wave Generator and Experiment Research of Signal Transmission Characteristic, China University of
Petroleum, Qing Dao, 2010.
[16] Y. Zhidan, W. Chunming, G. Yanfeng, et al., Design of a rotary valve orice for a continuous wave mud pulse generator, Precis. Eng. 41 (2015) 111118.
[17] M.A. Namuq, Simulation and Modeling of Pressure Pulse Propagation in Fluids Inside Drill Strings, TU Bergakademie Freiberg, Freiberg, 2012.
[18] M.A. Namuq, M. Reich, S. Bernstein, Continuous wavelet transformation: a novel approach for better detection of mud pulses, J. Petr. Sci. Eng. 110
(2013) 232242.
[19] P. Naseradinmousavi, C. Nataraj, Nonlinear mathematical modeling of buttery valves driven by solenoid actuators, Appl. Math. Model. 35 (5) (2011)
23242335.

[20] E. Lisowski, W. Czyzycki , J. Rajda, Three dimensional CFD analysis and experimental test of ow force acting on the spool of solenoid operated
directional control valve, Energy Convers. Manag. 70 (2013) 220229.
[21] X.G. Song, L. Wang, S.H. Baek, et al., Multidisciplinary optimization of a buttery valve, ISA Trans. 48 (3) (2009) 370377.
[22] M. Wannassi, F. Monnoyer, Numerical simulation of the ow through the blades of a swirl generator, Appl. Math. Model. 40 (2) (2016) 12471259.