You are on page 1of 13

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/348031369

Optimization of Flight Parameters for Petrel-L Underwater Glider

Article in IEEE Journal of Oceanic Engineering · December 2020


DOI: 10.1109/JOE.2020.3030573

CITATIONS READS
21 319

6 authors, including:

Shuxin Wang Ming Yang


Xiamen University Tianjin University
183 PUBLICATIONS 3,098 CITATIONS 19 PUBLICATIONS 188 CITATIONS

SEE PROFILE SEE PROFILE

Yanhui Wang
Tianjin University
101 PUBLICATIONS 967 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

An ERP approach to thematic hierarchies regarding grammatical objects of the Chinese verb Chi (eat) View project

Medical robotics View project

All content following this page was uploaded by Ming Yang on 12 October 2021.

The user has requested enhancement of the downloaded file.


This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

IEEE JOURNAL OF OCEANIC ENGINEERING 1

Peer-Reviewed Technical Communication


Optimization of Flight Parameters for Petrel-L Underwater Glider
Shuxin Wang, Ming Yang, Yanhui Wang , Member, IEEE, Shaoqiong Yang , Shiquan Lan, and Xinhai Zhang

Abstract—With recent developments in battery technology and low


power technology, biofouling, a process referring to the gradual accu-
mulation of organisms on underwater surfaces, has gained a foothold as
the primary adversary in long-duration underwater glider flights (C. D.
Haldeman, D. K. Aragon, T. Miles, S. M. Glenn, and A. G. Ramos, “Less-
ening biofouling on long-duration AUV flights: Behavior modifications and
lessons learned,” in Proc. OCEANS MTS/IEEE Conf., Monterey, CA, USA,
2016, pp. 1–8). Not only can biofouling increase the drag of the glider, but
also it can add the buoyancy discrepancies, which changes the flight charac-
teristics of the glider. Thus, it is necessary to adjust the flight parameters of
underwater gliders accordingly for long-duration flights. In this article,
the problems of a Petrel-L glider caused by biofouling are introduced.
Several critical parameters including the hydrodynamic coefficients, the
thermal expansion coefficient, the compressibility, and the net buoyancy
compensation of the glider are optimized by an inner penalty function
method based on the glider data and the dynamic model of Petrel-L. The Fig. 1. Biofouled Petrel-L glider.
effect of these parameters influenced by biofouling on the gliding range is
analyzed. The results indicate that the changed hydrodynamic coefficients
and net buoyancy compensation decrease the gliding range by 14.64% when
the hotel load, including the power of the control system and the sensors, characteristics throughout a mission. These changes, particularly in the
is 0.2 W. The gliding range loss can be decreased to 7.11% by adjusting hydrodynamics of the glider, result in larger volume, more drag and
the flight parameters. This article provides a reference of flight parameter added buoyancy discrepancies of the vehicle as well [1]. Thus, it is
adjustment for other types of biofouled underwater gliders.
increasingly evident that biofouling is a major factor limiting mission
Index Terms—Biofouling, flight parameter, optimization, underwater durations of the glider and transit distances it achieves [10].
glider. Marine biofouling is the undesirable growth of organisms on sub-
merged surfaces [11]. Any clean artificial substrata will be colonized
by bacteria, diatoms, and other microscopic unicellular eukaryotes
I. INTRODUCTION within hours after submersion [12]. Whether a glider is susceptible to
biofouling attachment depends on a number of environmental and op-

U NDERWATER gliders (UGs) are one of the most promising


ocean observation platforms, allowing the autonomous collec-
tion of long-term oceanographic data over long distances [2]. Driven by
erational factors. The most important ones among these are geographic
location, water temperature, mission duration, and operational depth
[10]. Some biofouling phenomena occurring to the gliders referred to
buoyancy, a UG can be equipped with various sensors to continuously
record physical, chemical, and biological parameters [3], [4]. Since the above have been reported in the literature. Seaglider 616 and Seaglider
concept of UG was first demonstrated [5], some typical UGs, such as the 516 deployed near the Hebrides and Brisbane, respectively, showed a
Slocum [6], the Seaglider [7], and the Spray [8], have been developed steep increase in the predicted drag coefficient toward the end of the
and widely applied in ocean observations for their outstanding cruising mission, the reason of which is the physical and reflective marine growth
ability [9]. However, the long duration and slow movement of gliders [13]. During the typical three-month missions in the Solomon Sea,
make them vulnerable to biofouling, which causes changes in flight Spray gliders were inevitably biofouled, particularly by barnacles [14].
After 160 days at sea in 2008, Slocum RU17 sank, approximately 200
miles west of the Azores, presumably caused primarily by a large animal
interaction. In April 2009, the glider RU 27 was once again launched off
Manuscript received February 6, 2020; revised July 21, 2020; accepted the New Jersey coast and headed across the Atlantic. After four months
October 2, 2020. This work was supported in part by the National Natural
Science Foundation of China under Grant 51722508, Grant 51721003, and Grant of flight in the summertime waters of the North Atlantic, it could no
11902219; in part by the National Key R&D Program of China under Grant longer maintain a heading in the intended direction of travel because
2016YFC0301100; and in part by the Natural Science Foundation of Tianjin of the biofouling [1]. In addition, the Petrel-L glider [15], designed by
City under Grant 18JCQNJC05100 and Grant 18JCJQJC46400. (Corresponding Tianjin University, Tianjin, China, in 2017, was biofouled by hydroids
author: Yanhui Wang.)
and covered by their mucus, as shown in Fig. 1 after traveling in the
Associate Editor: K. Asakawa.
The authors are with the Key Laboratory of Mechanism Theory and Equip- South China Sea for more than three months. Similarly, some flight
ment Design, Ministry of Education, School of Mechanical Engineering, Tianjin characteristics changed, including a reduction in gliding velocity and
University, Tianjin 300350, China, and also with the The Joint Laboratory of an increase in gliding time for one cycle. Moreover, the maximum
Ocean Observing and Detection, Pilot National Laboratory for Marine Sci- depth for one profile became smaller gradually in the sea trial when the
ence and Technology, Qingdao 266237, China (e-mail: shuxinw@tju.edu.cn;
yangming2019@126.com; yanhuiwang@tju.edu.cn; shaoqiongy@tju.edu.cn; target depth and other flight parameters were not adjusted. Thus, the
yxlx2010@163.com; 2018201107@tju.edu.cn). buoyancy of Petrel-L is deduced to be increased when it was biofouled.
Digital Object Identifier 10.1109/JOE.2020.3030573 The biofouling problem of gliders can be solved by coating its surface

0364-9059 © 2020 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

2 IEEE JOURNAL OF OCEANIC ENGINEERING

with antifouling paint [10] and adjusting the flight parameters [1]. The
former aims to reduce the biofouling and the latter can improve the
cruise duration when the flight characteristics are changed due to the
biofouling. This article focuses on the flight parameter adjustment of
the UG, which is based on the identification of the flight parameters.
In terms of identification of flight parameters, Graver et al. [16], [17]
identified the model parameters to match the steady glides in new flight
data from the Slocum glider, without considering the variation of the
net buoyancy. Ma et al. [18] optimized the hydrodynamic parameters
by integrating electromagnetic velocity sensor on the glider, which can
exhibit dynamic behaviors of the glider more accurately. Williams et al.
[19] developed and validated a new model for predicting the behaviors Fig. 2. Schematic representation of the glider moving in a vertical plane, the
of an ocean glider by a specially formulated parameter-identification forces involved, and the definition of angles. (The forces working on the glider
scheme. Merckelbach et al. [20] identified the parameters of drag include buoyancy FB , gravity Fg , lift L, and drag D. The angle of attack α is
exactly the difference between the gliding angle γ and the pitch angle θ.)
coefficient, glider volume (at atmospheric pressure), and hull compress-
ibility by applying the model to data gathered during the Gulf of Lions
2008 experiment. Todd et al. [21] described how glider-based Doppler pitch angle, and roll angle) are measured with a TCM3 sensor (PNI
current measurement could be used to infer glider flight characteristics Sensor Corporation, Santa Rosa, CA, USA)2 . The hulls of Petrel-L
and, in turn, corrected vertically averaged current estimates to account glider are made of 6061T6 aluminum alloy. The dormancy strategy
for accumulated biofouling or other changes in glider flight during of the control system is optional, and the sampling frequencies of the
a mission. Merckelbach et al. [22] utilized data collected from an sensors can be adjusted according to different missions. Thus, the hotel
electromagnetic current meter mounted inside a microstructure package load is referred to as a variable in the following analysis.
to assess flight characteristics of the glider.
Although the above-mentioned investigations have provided some
methods to identify the flight parameters of the biofouled UG, the
B. Dynamic Model of a Petrel-L Glider for Planar Flight
following problems still exist. The dynamic model of the UG has been introduced in previous
1) The flight parameters influenced by biofouling are not entirely literature [23]–[26]. Petrel-L glider is designed to have separate control
investigated in the previous work. Not only are the lift coefficient algorithms and actuators for controlling the lateral dynamics and verti-
and drag coefficients including the parasite drag coefficient and cal dynamics. In this study, the vertical plane dynamics are controlled
induced drag coefficient influenced by biofouling, but the net by the movable mass. Here, the lateral dynamics can be ignored since
buoyancy, the compressibility, and the thermal expansion coeffi- a fixed back rudder is adopted to make the vehicle glide in a straight
cient of the glider may also be varied with the biofouling, which motion in a vertical plane and thus stabilize the dynamics [15]. Thus, it is
has specific influences on the flight characteristics. useful to study the vertical plane dynamics, which represent the actual
2) The effect of the changed flight parameters of the biofouled glider glider operation. Besides, it is reasonable to neglect the acceleration
on the gliding range and the effect of flight parameter adjustment terms for motions with time scales greater than required for the glider
have not been researched in detail. to travel several body lengths. At a speed of 0.5 m/s, the glider travels
This article is organized as follows. Section II introduces the glider more than ten body lengths (30 m) in 1 min. Therefore, when gliding in
specification and the dynamic model of Petrel-L. Section III describes motions with time scales larger than 1 min, the glider can be considered
the proposed method to identify the flight parameters and the verifica- to be in quasi-steady flight [20].
tion of the optimization results. In Section IV, the effects of the flight A schematic representation of the gilder, the forces that work on it,
parameters on the gliding range are analyzed, and the flight parameters and the definition of rotation angles are shown in Fig. 2. As schematized,
are adjusted to reduce the impact of their changes. Besides, future plans all forces are considered to originate from the center of buoyancy for
and possible applications of the results are discussed. Finally, Section V the sake of clarity, which is permissible because a torque balance is not
concludes this article. considered herein. Also note that the glide and pitch angles are negative
when the glider points down. The vertical and horizontal force balances
II. GLIDER MODEL then read

A. Glider Specification FB + cos γL + sin γD − Fg = 0 (1a)

This study adopts the Petrel-L glider designed by Tianjin University. cos γD − sin γF = 0 (1b)
Details and specifications of this glider can be found in [15]. This
respectively, where FB is the buoyancy force, Fg is the force due to
Section merely introduces the details relevant to this work.
gravity, L is the lift force, D is the drag force, and γ is the gliding angle,
The Petrel-L is designed for the observation of large-scale ocean
defined as the sum of the pitch angle θ and the angle of attack α.
phenomena and has a horizontal gliding range larger than 3000 km.
The force due to gravity is given by
Petrel-L can perform over 700 gliding cycles in 1 mission, and the
averaged horizontal gliding range for one 1000-m-deep cycle is about Fg = mg g (2)
5 km. The propulsion of glider is a buoyancy engine, which has a
maximum volume change of 1.4 L. The conductivity, temperature, and where mg is the mass of the glider and g is the acceleration due to
depth (CTD) sensor onboard is a Seabird Glider Payload CTD1 , the gravity. The net buoyancy force is given by
sampling frequency of which can be adjusted. The attitudes (heading, FB = gρ {Vg [1 − εP + αT (T − T0 )] + ΔVbp } + ΔB (3)

1 https://www.seabird.com/moving-platform/glider-payload-ctd-gpctd 2 http://www.pnicorp.com/products/all/tcm-legacy

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

WANG et al.: OPTIMIZATION OF FLIGHT PARAMETERS FOR PETREL-L UNDERWATER GLIDER 3

where ρ is the in situ density, Vg is the glider volume at atmospheric


pressure, ε is the compressibility of the hull, P is the water pressure, αT
is the thermal expansion coefficient, T is the water temperature, T0 is
a reference water temperature, ΔVbp is the buoyancy change resulting
from the buoyancy engine, and B is the compensation value of the
net buoyancy. When the glider is at the seawater surface, the force due
to gravity is equal to ρVg g. The drag and lift forces are given by
1
D= ρCD AU 2 (4)
2
1
L = ρCL AU 2 (5)
2
where CD and CL are the coefficients of drag and lift, respectively, A
Fig. 3. Moving trajectory of Petrel-L.
is the cross-sectional area of the glider, and U is the gliding velocity
through the water along the glide path.
The total lift is modeled as the sum of the lift generated by the hull and
that generated by the wings. Under normal flight conditions, the angle
of attack is generally small. For small angles of attack, lift coefficient,
and drag coefficient can be parameterized as
C L = KL α (6)
2
CD = KD0 + KD α (7)
where KL is the lift coefficient, KD0 is the parasite drag coefficient, and
KD is the induced drag coefficient.
Substituting (6) and (7) into (5) and (4), respectively, yields
1  
D= ρ KD0 + KD α2 AU 2 (8)
2
1
L = ρKL αAU 2 . (9)
2
The vertical velocity Uv can be calculated by U as Fig. 4. Glider data for one gliding profile.

Uv = U sin (α − θ) . (10)
Solving (1a), (1b), (2), (3), (8), (9), and (10), the vertical velocity
flight parameters ε, αT , and B of Petrel-L may be changed when
can be obtained as shown in (11) and (12) at the bottom of the page.
the glider operates in the seawater, especially when it is biofouled. In
In (11), the pitch angle θ can be measured by the TCM3 sensor,
addition, the biofouling changes the hydrodynamic shape of Petrel-L,
whereas the angle of attack cannot be measured. Thus, by solving
which may change the values of KL , KD0 , and KD .
(11), the angle of attack α expressed by pitch angle θ can be deduced.
However, it is difficult to solve (11) because it involves the quadratic
and trigonometric terms of α. In this study, the trigonometric term on A. Data Set: South China Sea 2019
the right-hand side of (11) is transformed into a series and the complex
expression can be solved. Finally, by submitting this expression into To self-calibrate the model and assess its performance, the model
(12), the vertical velocity can be expressed by the flight parameters was applied to data obtained by Petrel-L during the South China Sea
including ε, αT , B, KL , KD0 , and KD . 2019 measurement campaign. The Petrel-L was deployed in April 2019
and successfully recovered in September 2019, having traveled over
3000 km and over 700 1000-m-deep gliding cycles, as shown in Fig. 3.
III. IDENTIFICATION OF THE FLIGHT PARAMETERS The glider data of Petrel-L for one gliding cycle can be seen in Fig. 4.
To evaluate the vertical velocity with (11) and (12), several parame- The temperature data and density data measured by the CTD sensor
ters need to be known. The sensor suite of the glider provides numerical are shown in Fig. 5, from which we can see that there is an eddy in the
input for θ, ΔVbp , T, P, and ρ. The specific parameters A, Vg , B, and density data around day 80.
mg of the glider can be measured directly, and the flight parameters ε, Only the 1000-m-deep gliding cycles are considered in the identifi-
αT , KL , KD0 , and KD can also be measured by pressure test and towing cation of the flight parameters. Before identifying the flight parameters
tank test. However, the compressibility, thermal expansion coefficient, with the glider data, the following conditions also need to be met to
and the net buoyancy of the fouling organisms are unknown. Thus, the ensure the steady-state motion.

CL KL α
= = cot γ = cot (α − θ) (11)
CD KD0 + KD α2

2 {mg g − gρ {Vg [1 − εP + αT (T − T0 )] + ΔVbp } − ΔB} sin3 (α − θ)
Uv = (12)
ρ (KD0 + KD α2 ) A

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

4 IEEE JOURNAL OF OCEANIC ENGINEERING

Fig. 5. (a) Temperature and (b) density profiles.

3) To reduce the influence of the current, the data points close to


the surface where nonlinearities are significant are abandoned.
4) When the pitch angles and headings are 5o larger than the set
value, the relevant data points are removed.
5) To avoid transient conditions, the data points where accelerated
velocity is larger than 0.01 m/s2 and angular velocity of pitch
angle, roll angle, and heading larger than 1°/s are removed.

B. Optimization
In this article, the flight parameters are optimized based on the glider
data. The optimization is established as follows.
1) Optimized Variables: The flight parameters, including ε, αT ,
B, KL , KD0 , and KD , are selected as variables to be optimized, which
have an obvious influence on the flight characteristics of the UG.
2) Objective Function: The gliding velocity, a main flight char-
acteristic of UGs, is achieved by inputting the flight parameters into the
proposed model, the variation of which influences the gliding velocity
by changing the net buoyancy and the hydrodynamic force of the UG.
However, the actual gliding velocity cannot be obtained only with the
glider data, because the angle of attack cannot be measured directly
by the conventional sensors carried by the Petrel-L. Thus, the vertical
velocity is involved in the objective function, which can be measured
by the pressure sensor and deduced by (11) and (12). In this article, the
objective function F is defined as

N
F = [Uv (ε, αT , ΔB, KL , KD0 , KD ) − Ug ]2 (13)
i=1

where N is the number of data points used in the optimization process,


and Ug is the vertical velocity calculated by the pressure sensor data.
3) Constraint Conditions: Considering some practical condi-
tions, some constraints need to be satisfied, which are described as
follows.
Fig. 6. Procedure of the optimization solution. a) Side constraints: Each optimized variable needs to have a
feasible region in which the optimum value can be achieved. Taking
reality into account, the scope of the imported flight parameters needs
to be wide enough to get a correct optimization result. Hence, the side
1) All aborted gliding cycles where Petrel-L climbs to the surface constraints of the six imported flight parameters can be expressed as
are removed.
2) The steady pitch angle, roll angle, and net buoyancy are required g1 : 1 × 10−10 Pa−1 ≤ ε ≤ 3 × 10−10 Pa−1 (14)
for steady-state motion. Thus, the time-series data when the −5o −1 −5o −1
g2 : 1 × 10 C ≤ αT ≤ 9 × 10 C (15)
buoyancy adjustment and attitude adjustment units are working
are removed. g3 : −3 N ≤ ΔB ≤ 3 N (16)

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

WANG et al.: OPTIMIZATION OF FLIGHT PARAMETERS FOR PETREL-L UNDERWATER GLIDER 5

Fig. 7. Optimization results. (a) Variation trend oftrend of the lift coefficient. (b) Variation trend of the parasite drag coefficient. (c) Variation trend of the induced
drag coefficient. (d) Variation trend of the compressibility. (e) Variation trend of the thermal expansion coefficient. (f) Variation trend of the the compensation value
of net buoyancy.

c) Other constraint: For the solution of (11) in the real number


field, one must have

g9 : (KL tan γ)2 − 4KD KD0 ≥ 0. (22)

4) Solution Algorithm: In this article, the optimization of flight


parameters is a linear constraint issue involving some inequality con-
straints. In this study, the inner penalty function method (IPFM) [27]
with high precision and efficiency is applied to solve this problem
effectively. The procedure of the optimization solution is shown in
Fig. 6, and the definition of the constraint equation gu (X) and the design
variables X are shown in the following equations:
Fig. 8. Recovery of biofouled Petrel-L (profile 455).
gu (X) = [g1 g2 g3 g4 g5 g6 g7 g8 g9 ] (23)

g4 : 0.3 ≤ KL ≤ 0.7 (17) X = [KL KD0 KD ε αT ΔB] . (24)

g5 : 0.2 ≤ KD0 ≤ 0.5 (18) 5) Optimization Results: As shown in Fig. 7, the flight param-
g6 : 0.006 ≤ KD ≤ 0.012. (19) eters, including ε, αT , B, KL , KD0 , and KD , are optimized. What
draws special attention is that to verify the effect of biofouling on the
b) Net buoyancy constraints: Petrel-L is driven by net buoy- hydrodynamic force, the glider was recovered provisionally in profile
ancy. Thus, in the descending process of the glider, the force due to 455 after the flight of 87 days to clean the biofouling, as shown in Fig. 8.
gravity should be larger than the buoyancy, whereas the buoyancy is Also, to verify the identification of B, the payload of 150 g is added
larger in the ascending process. Therefore, some constraints about net on the Petrel-L in profile 36. Thus, the obvious change in B, KL , KD0 ,
buoyancy need to be met in the optimization and KD can be found in Fig. 7. It can be seen from Fig. 7 that some
flight parameters including B, KL , KD0 , and KD increased linearly
g7 : Fg − FB ≥ 0 (when θ ≤ 0) (20)
with biofouling degree and other flight parameters including ε and αT
g8 : Fg − FB ≤ 0 (when θ ≥ 0). (21) varied little.

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

6 IEEE JOURNAL OF OCEANIC ENGINEERING

TABLE I
FLIGHT PARAMETERS IN TYPICAL PROFILES

Fig. 10. Pressure test.

TABLE III
COMPARISON OF IDENTIFICATION RESULTS AND SIMULATION RESULTS

Fig. 9. Volume change of the pressure hull. (V is the volume change of the
pressure hull, T is the temperature difference, and P is the pressure. The points
are the simulation results, and the line is fitted by the points.)

TABLE II
COMPARISON OF IDENTIFICATION RESULTS AND SIMULATION RESULTS
correspond well to the simulation results. To validate the simulation
results, a pressure test is performed to measure the volume variation of
the pressure hull, the principle of which is shown in Fig. 10. The water
inside the pressure hull will flow into the bladder when the hydrostatic
pressure against the pressure hull increases. Thus, the volume change of
the pressure hull can be obtained by measuring the water volume in the
bladder, which is 96 mL under the hydrostatic pressure of 10 MPa. In the
pressure test, the environment temperature and water temperature in the
To decrease the error of the identification, the averaged values are pressure tank are 28 °C and 12 °C, respectively, which are imported into
calculated as the flight parameters, and ε and αT are 2.1874 × 10–10 1/Pa the simulation, and the volume variation is calculated as 99.8 mL in the
and 2.8364 × 10–5 1/°C, respectively. The parameters in typical profiles simulation. The error between the pressure test result and the simulation
are selected, as shown in Table I, from which it can be calculated that result is 4%, which verifies the efficiency of the simulation.
the decreases in B are 1.4398 N when the payload is added. Besides, b) Hydrodynamic coefficient: When the fluid flows past UGs,
the increase in B is 0.5949 N from profile 37 to profile 455, and the viscous turbulent flow happens at the fluid vehicle boundary layer,
increase rates of KL , KD0 , and KD are 23.44%, 31.7%, and 33.2% from which produces the lift and drag [28]. The Reynolds-averaged Navier–
profile 20 to profile 455, respectively. Stokes equations are efficient for hydrodynamic simulations of UGs.
6) Verification of the Optimization Results: In this article, the computational fluid dynamics (CFDs) software,
a) Compressibility and thermal expansion coefficient: The ANSYS Fluent, is used to calculate the hydrodynamic coefficients.
pressure hull takes 93% of the total volume of Petrel-L and the pay- Because of the low Reynolds number, the shear stress transport (SST)
loads outside the pressure hull are incompressible. Thus, the volume k-ε turbulence model is applied, which combines the advantages of the
change of the payloads outside the pressure hull is negligible. To verify k-ε turbulence model and the k-ω turbulence model.
the compressibility and thermal expansion coefficient of Petrel-L, the Before computing the hydrodynamic coefficients, the computer-
finite element analysis code ANSYS Workbench is utilized to analyze aided design (CAD) model is established and imported into the ICEM
the deformation and volume variation of the pressure hull that takes CFD software to divide the mesh. The computation domain is cylin-
the temperature and hydraulic pressure changes into account. The drical, which is selected as 8Lglider × 20Dglider × 20Dglider. Then,
volume change of the pressure hull is shown in Fig. 9 and the com- the mesh is imported into the ANSYS Fluent software to calculate
pressibility and thermal expansion coefficient can be fitted by the data the hydrodynamic coefficients. The results of the CFD simulation
points of the simulation, which are 2.1254 × 10–10 1/Pa and 2.8152 × are shown in Fig. 11 and Table III, from which we can see that the
10–5 1/°C, respectively. It can be seen from Table II that the identifica- identification results have good consistency with simulation results.
tion results of the compressibility and the thermal expansion coefficient What needs to be noted is that the drag coefficient KD0 decreases with

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

WANG et al.: OPTIMIZATION OF FLIGHT PARAMETERS FOR PETREL-L UNDERWATER GLIDER 7

Fig. 11. CFD simulation results. (a) Simulation results of the lift coefficient. (b) Simulation results of the drag coefficient.

importing the angle of attack into (6) and (7), respectively. Thus, the
variation trend of the lift coefficient CL and drag coefficient CD with the
pitch angle can be obtained, as shown in Fig. 13 (a) and (b). The change
rates of CL and CD from profile 20 to profile 455 are shown in Fig. 13(c),
from which we can see that the change rate of CL and CD reaches
over 30% and it becomes larger with the decrease of the pitch angle.
Fig. 13(d) shows that the change of the hydrodynamic coefficients has
little influence on the lift-drag ratio at a given pitch angle, meaning that
the gliding angle changes little in profile 20 and profile 455 at a given
pitch angle according to (11). Thus, the gliding range for one gliding
cycle changes little when the hydrodynamic coefficients vary with time.
However, by importing the hydrodynamic coefficients of profile 20 and
profile 455 respectively into the glider model, the gliding velocity can
be obtained, which decreases with the hydrodynamic coefficients, as
shown in Fig. 13(e). Therefore, the gliding time for one gliding cycle
Fig. 12. Variation of KD0 with the velocity. becomes longer when the net buoyancy is not adjusted and the energy
consumption of control system and sensors related to time increases,
the velocity in the simulation. According to the gliding velocity in which lowers the gliding efficiency and decreases the total gliding
the sea trial (0.25–0.5 m/s), the drag coefficient KD0 is calculated, as range. It explains why the gliding range of Petrel-L for one gliding cycle
shown in Fig. 12. However, the drag coefficient KD0 in the glider model changes little and the gliding time of one cycle increases gradually in
is referred to as a constant, for the convenience of simulation in the sea trial.
current study, as shown in (7). Thus, the error of KD0 is a little larger 2) Net Buoyancy Compensation: When the net buoyancy com-
than that of other hydrodynamic coefficients. pensation decreases with time and the buoyancy change ΔVbp is not
c) Net buoyancy compensation: The identification of net adjusted, the gliding velocity of diving phase increases and that of
buoyancy compensation is verified by the payload adjustment in profile climbing phase decreases, which may have a certain influence on the
36. The masses of the added payload are 150 g, and the variety of the gliding range.
net buoyancy compensations of identification is 1.4398 N, the error of Taken together, Fig. 14 shows the effects of hydrodynamic coeffi-
which is 4.01%. cients and net buoyancy compensation on gliding velocity and gliding
trajectory with unadjusted parameters. From Fig. 14 and Table IV,
IV. RESULTS OF FLIGHT PARAMETER ADJUSTMENT we can see that the change in the hydrodynamic coefficients has a
AND DISCUSSIONS larger influence on the averaged gliding velocity and time. Since the
As been described above, some flight parameters of Petrel-L, includ- net buoyancy compensation has an opposite effect on the diving velocity
ing hydrodynamic coefficients and net buoyancy compensation, vary and climbing velocity is opposite, it has little influence on the gliding
with time. Therefore, the flight parameters, including the pitch angle time. The gliding time for one gliding cycle increases by 17.76% when
θ and the buoyancy change ΔVbp resulting from the buoyancy engine, the hydrodynamic coefficients and net buoyancy compensation are
need to be adjusted for the highest gliding efficiency and the longest changed, which increases the energy consumption of glider for one
gliding range. Before the adjustment of flight parameters is analyzed, gliding cycle.
the effect of changed flight parameters on the motion characteristics of
Petrel-L is studied.
B. Flight Parameter Adjustment

A. Influence Analysis of Changed Flight Parameters As analyzed above, when the hydrodynamic coefficients of the
biofouled glider including KL , KD0 , and KD increase, the gliding time
1) Hydrodynamic Coefficients: The angle of attack for a given for one cycle increases, which increases the energy consumption of
pitch angle can be calculated by the hydrodynamic coefficients (KL , the measurement unit and control unit and decreases the total gliding
KD0 and KD ) shown in (11), and CL and CD can be obtained by range. Thus, the flight parameters, including θ and ΔVbp , need to be

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

8 IEEE JOURNAL OF OCEANIC ENGINEERING

Fig. 13. Comparison of (a) lift coefficient, (b) drag coefficient, (c) change rate of CL and CD , (d) lift drag ratio, and (e) gliding velocity. [In (e), the red surface
is simulated with the hydrodynamic coefficient of profile 1, and the green one is simulated with that of profile 455.]

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

WANG et al.: OPTIMIZATION OF FLIGHT PARAMETERS FOR PETREL-L UNDERWATER GLIDER 9

TABLE IV
CHANGE RATE OF GLIDING VELOCITY AND GLIDING TIME

and climbing phases decrease and the buoyancy change resulting from
the buoyancy engine ΔVbp increases with the gradual accumulation of
biofouling. Fig. 15(d) shows the variation trend of the ratio of gliding
range to the energy consumption for one gliding cycle.
Fig. 16 shows the effect of the changed hydrodynamic coefficients
and net buoyancy compensation on the gliding range when the flight
parameters are not adjusted. It can be seen that the reduction rate is
larger when the hotel load is smaller. When the hotel load is 0.2 W, the
reduction rate of profile 455 reaches 14.64%. Fig. 17 shows the effect
of the flight parameter adjustment on the gliding range and the trend
of increase rate is similar to that of the reduction rate in Fig. 16. The
flight parameter adjustment can increase the gliding range by 7.53%
and the gliding range loss can be decreased to 7.11% when the hotel
load is 0.2 W in profile 455. The increase rate turns to be larger with
the gradual accumulation of biofouling.

C. Plans and Possible Applications


With the optimization of a Petrel-L UG in its carried battery and
energy consumption, the cruising ability will be further promoted and
the effect of the biofouling problem on the gliding range will become
more and more obvious. Thus, adjusting the flight parameters in real
time will be more important for a longer gliding range. Thus, a plan of
flight parameter adjustment will be drawn for some long-deployment
glider missions. The glider data will be analyzed with the method in
this article for every 20 gliding cycles to optimize the flight parameters
ε, αT , B, KL , KD0 , and KD . According to the optimization, the flight
parameters including θ and ΔVbp will be adjusted autonomously to
weaken the influence of the biofouling and increase the gliding range.
In turn, the rate of biofouling can be detected by the variation trend of
the flight parameter B, as shown in Fig. 7(f).
The optimization method and the sea trial results provide a reference
for other UGs performing the long-deployment glider missions. The
Fig. 14. Comparison of (a) gliding velocities and (b) gliding trajectories. results in Section IV-B provide a direction for flight adjustment of other
gliders to weaken the effect of biofouling. As introduced in Section I,
the biofouling of the glider depends on a number of environmental
adjusted accordingly to reduce the effect of the changed hydrodynamic and operational factors. Therefore, other gliders may have a differ-
coefficients. In this article, the optimization method in [15] is adopted ent biofouling degree from that of Petrel-L in this article. Thus, the
to obtain the adjusted flight parameters. In the optimization, the angles specific adjustment of flight parameters for other gliders depends on
of attack and the buoyancy changes resulting from the buoyancy engine the optimization results with the method proposed in this article, with
are selected as the optimized variables, and the objective function is the consideration of different flight parameters.
ratio of the gliding range to the energy consumption for one profile.
The profiles 20, 165, 310, and 455 are selected to analyze the varia-
tion trend of the optimized flight parameters. As shown in Fig. 15, the
V. CONCLUSION
variation trend of the adjusted flight parameters is described. The x-axis In this study, a dynamic model of a Petrel-L glider for planar flight
in Fig. 15 represents the hotel load, the range of which is 0.2–4 W. It can is established and the flight parameters are identified based on the
be seen from Fig. 15(a)–(c) that the angles of attack α in both diving filtered glider data by the inner penalty function method, the correctness

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

10 IEEE JOURNAL OF OCEANIC ENGINEERING

Fig. 15. Optimization results of flight parameter.

Fig. 16. Reduction rate of gliding range. Fig. 17. Increase rate of gliding range.

of which is verified by pressure test, CFD simulation, and payload larger hydrodynamic coefficients. The following conclusions can be
adjustment in the sea trial. Then, the variation trend and influence of the drawn.
flight parameters changing with the severity of biofouling are analyzed. 1) The optimization in this article that combines the dynamic model
Finally, the adjusted flight parameters including the pitch angle θ and for planar flight and the glider data is effective to identify the
the buoyancy change ΔVbp are proposed for reducing the effect of the changed flight parameters of the biofouled glider.

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

WANG et al.: OPTIMIZATION OF FLIGHT PARAMETERS FOR PETREL-L UNDERWATER GLIDER 11

2) The gradual accumulation of biofouling has little influence on [14] R. E. Davis, W. S. Kessler, and J. T. Sherman, “Gliders measure western
the compressibility ε and thermal expansion coefficient αT . boundary current transport from the south pacific to the equator,” J. Phys.
However, after 87 days’ flight, the hydrodynamic coefficients Oceanogr., vol. 42, no. 11, pp. 2001–2013, 2012.
[15] M. Yang, Y. Wang, S. Wang, S. Yang, Y. Song, and L. Zhang, “Motion pa-
KL , KD0 , and KD increase by 23.44%, 31.7%, and 33.2%, rameter optimization for gliding strategy analysis of underwater gliders,”
respectively, and the net buoyancy compensation B increases Ocean Eng., vol. 191, 2019, Art. no. 106502.
by 0.5949 N. [16] J. G. Graver, R. Bachmayer, and N. E. Leonard, “Underwater glider model
3) The changed flight parameters increase the gliding time of a parameter identification,” presented at the 13th Int. Symp. Unmanned
Untethered Submersible Technol., Durham, NH, USA, Aug. 2003.
biofouled glider by 17.76% for one gliding cycle, which increases [17] J. G. Graver, “Underwater gliders: Dynamics, control and design,” Dept.
the energy consumption and decreases the gliding range by Mech. Aerosp. Eng., Princeton Univ., Princeton, NJ, USA, 2005.
14.64% when the hotel load is 0.2 W. [18] W. Ma, Y. Wang, S. Wang, G. Li, and S. Yang, “Optimization of hydro-
4) The angle of attack α and the buoyancy change resulting from dynamic parameters for underwater glider based on the electromagnetic
the buoyancy engine ΔVbp need to be adjusted to be smaller and velocity sensor,” Proc. Inst. Mech. Eng., C, J. Mech. Eng. Sci., vol. 233,
no. 14, pp. 5019–5032, 2019.
larger, respectively, for the largest gliding range with the gradual [19] C. D. Williams, R. Bachmayer, and B. de Young, Toward Improved Predic-
accumulation of biofouling. tions of the Performance of Ocean Gliders. Lee, NH, USA: Autonomous
5) The flight parameter adjustment of the glider is an efficient Undersea Systems Institute, 2007.
measure to reduce the effect of the biofouling on gliding range, [20] L. Merckelbach, D. Smeed, and G. Griffiths, “Vertical water velocities
from underwater gliders,” J. Atmos. Ocean. Technol., vol. 27, no. 3,
which decreases the gliding range loss to 7.11% for Petrel-L pp. 547–563, 2010.
when the hotel load is 0.2 W. The increase rate is more obvious [21] R. E. Todd, D. L. Rudnick, J. T. Sherman, W. B. Owens, and L.
with the gradual accumulation of biofouling and the decrease of George, “Absolute velocity estimates from autonomous underwater gliders
the hotel load. equipped with Doppler current profilers,” J. Atmos. Ocean. Technol.,
vol. 34, no. 2, pp. 309–333, 2017.
Future works will focus on the strategy for reducing the biofouling
[22] L. Merckelbach, A. Berger, G. Krahmann, M. Dengler, and J. R. Carpenter,
of the glider and the flight parameter variation of longer duration gliders “A dynamic flight model for slocum gliders and implications for turbulence
needs to be studied to increase the gliding range. microstructure measurements,” J. Atmos. Ocean. Technol., vol. 36, no. 2,
pp. 281–296, 2019.
[23] Y. Singh, S. K. Bhattacharyya, and V. G. Idichandy, “CFD approach
ACKNOWLEDGMENT to modelling, hydrodynamic analysis and motion characteristics of a
laboratory underwater glider with experimental results,” J. Ocean Eng.
The authors would like to express their sincere thanks to L. Ma for Sci., vol. 2, no. 2, pp. 90–119, 2017.
her help in revising the grammar. [24] P. Yu, T. Wang, H. Zhou, and C. Shen, “Dynamic modeling and three-
dimensional motion simulation of a disk type underwater glider,” Int. J.
Nav. Arch. Ocean, vol. 10, no. 3, pp. 318–328, 2018.
REFERENCES [25] Y. Yang, Y. Liu, Y. Wang, H. Zhang, and L. Zhang, “Dynamic modeling
and motion control strategy for deep-sea hybrid-driven underwater gliders
[1] C. D. Haldeman, D. K. Aragon, T. Miles, S. M. Glenn, and A. G.
considering hull deformation and seawater density variation,” Ocean Eng.,
Ramos, “Lessening biofouling on long-duration AUV flights: Behavior
vol. 143, pp. 66–78, 2017.
modifications and lessons learned,” in Proc. MTS/IEEE OCEANS Conf.,
[26] N. A. A. Hussain, M. R. Arshad, and R. Mohd-Mokhtar, “Underwater
Monterey, 2016, pp. 1–8.
glider modelling and analysis for net buoyancy, depth and pitch angle
[2] D. L. Rudnick, R. E. Davis, C. C. Eriksen, D. M. Fratantoni, and M.
control,” Ocean Eng., vol. 38, no. 16, pp. 1782–1791, 2011.
J. Perry, “Underwater gliders for ocean research,” Mar. Technol. Soc. J.,
[27] Y. G. Evtushenko, Numerical Optimization Techniques. New York, NY,
vol. 38, no. 2, pp. 73–84, 2004.
USA: Optimization Software, Publications Division, 1985.
[3] D. L. Rudnick, “Ocean research enabled by underwater gliders,” Annu.
[28] S. Zhang, J. Yu, A. Zhang, and F. Zhang, “Spiraling motion of underwa-
Rev. Mar. Sci., vol. 8, no. 1, pp. 519–541, 2016.
ter gliders: Modeling, analysis, and experimental results,” Ocean Eng.,
[4] R. Bachmayer, N.E. Leonard, J. Graver, E. Fiorelli, P. Bhatta, and D. Paley,
vol. 60, pp. 1–13, 2013.
“Underwater gliders: Recent developments and future applications,” in
Proc. Int. Symp. Underwater Technol., 2004, pp. 195–200.
[5] H. Stommel, “The SLOCUM mission,” Oceanography, vol. 2, no. 1,
pp. 22–25, 1989.
[6] D. C. Webb, P. J. Simonetti, and C. P. Jones, “SLOCUM: An underwater
glider propelled by environmental energy,” IEEE J. Ocean. Eng., vol. 26,
Shuxin Wang is the Vice-President of Tianjin Uni-
no. 4, pp. 447–452, Oct. 2001.
versity, Tianjin, China. He is also a member of the
[7] C. C. Eriksen et al., “Seaglider: A long-range autonomous underwater
Technical Committee for Multibody Dynamics with
vehicle for oceanographic research,” IEEE J. Ocean. Eng., vol. 26, no. 4,
International Federation for the Promotion of Mech-
pp. 424–436, Oct. 2001.
anism and Machine Science. He is the Director of
[8] J. Sherman, R. E. Davis, W. B. Owens, and J. Valdes, “The autonomous un-
Ministry of Education, Key Laboratory of Mechanism
derwater glider “Spray”,” IEEE J. Ocean. Eng., vol. 26, no. 4, pp. 437–446,
Theory and Equipment Design established at Tianjin
Oct. 2001.
University, and the Director of the National Research
[9] S. A. Jenkins et al., “Underwater glider system study,” Office Nav. Res.,
Foundation Innovation Group of High-end Equip-
Code 321 OE, Ocean Eng. Mar. Syst., Arlington, VA, USA, Scripps Inst.
ment Mechanism Theory and Technology Base. He
Oceanogr. Tech. Rep. 53, 2003.
has authored or coauthored more than 100 high-level
[10] H. Lobe, C. Haldeman, and S. M. Glenn, “Clearsignal coating controls
academic papers, 100 plus of them included in SCI/EI. He has applied for more
biofouling on the Rutgers glider crossing,” Sea Technol., vol. 5, no. 51,
than 150 national invention patents, among which 90 plus are authorized. He
pp. 31–36, 2010.
currently has 17 PCT applications, three patents authorized by the United States,
[11] M. Wahl, “Marine epibiosis. I. Fouling and antifouling: Some basic
and one patent authorized by Europe. His research interests mainly involve the
aspects,” Mar. Ecol. Prog. Ser., vol. 58, pp. 175–189, 1989.
intelligent robotics (minimally invasive surgical robot, underwater glider, etc.)
[12] M. Salta, J. A. Wharton, Y. Blache, K. R. Stokes, and J.-F. Briand,
and mechanical system dynamics and control. He has supervised more than
“Marine biofilms on artificial surfaces: Structure and dynamics,” Environ.
40 projects of international cooperation and at national and ministry levels.
Microbiol., vol. 15, no. 11, pp. 2879–2893, 2013, doi: 10.1111/1462-
Mr. Wang won a second prize of National Scientific and Technological
2920.12186.
Progress Award, a second prize of National Technological Innovation, and six
[13] E. Anderlini, C. Harris, A.B. Phillips, A. L. Lopez, M. Woo, and G.
other awards at the ministry level. He was the winner of the National Science
Thomas, “Towards autonomy: A recommender system for the determi-
Fund for Distinguished Young Scholars of China, a Yangtze River Scholar of the
nation of trim and flight parameters for seagliders,” Ocean Eng., vol. 189,
Ministry of Education, and a Technological Leading Talent listed in the Third
2019, Art. no. 106338.
Batch of National “Ten-Thousand Talents Plan.”

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.

12 IEEE JOURNAL OF OCEANIC ENGINEERING

Ming Yang received the B.S. degree from Tianjin Shaoqiong Yang received the combined master’s and
University, Tianjin, China, in 2015, where he is cur- Ph.D. degrees in fluid mechanics from the Depart-
rently working toward the Ph.D. degree with the ment of Mechanics, School of Mechanical Engineer-
School of Mechanical Engineering, both in electrical ing, Tianjin University, Tianjin, China, in 2015. He
engineering. studied for one year at the University of Nottingham,
His research interests include overall design, mo- Nottingham, U.K., from 2013 to 2014, under the
tion parameters optimization, and drag reduction supervision of Prof. K.-S. Choi.
strategies for underwater gliders. From 2015 to 2017, he was a Research Fellow at
Nanyang Technological University, Singapore. Since
2017, he has been with the Underwater Robot Team,
School of Mechanical Engineering, Tianjin Univer-
sity. He is currently an Associate Professor, Lecturer and the Master’s Tutor with
the School of Mechanical Engineering, Tianjin University. His research interests
include the experimental fluid mechanics on flow control for turbulence and the
hydrodynamic shape optimization for autonomous underwater vehicle/glider
and their drag reduction strategies.
Dr. Yang was a recipient of the Ministry of Education Academic Newcomer
Yanhui Wang (Member, IEEE) received the B.S., Award, China, in 2012.
M.S., and Ph.D. degrees in mechanical engineering
from Tianjin University, Tianjin, China, in 2001,
2004, and 2008 respectively. Shiquan Lan received the B.S. degree from Qingdao
He is currently the Ph.D. graduates’ supervisor. He University, Qingdao, China, in 2010, and the M.S.
is also the Deputy Director of the Marine Equipment degree from the School of Mechanical Engineering,
Technology Engineering Center of Tianjin and also Tianjin University, Tianjin, China, in 2013, both in
the Director of the Marine Observation Branch of mechanical engineering.
Chinese Oceanology and Limnology Institute. His research interests include the control strategy
Mr. Wang is the Editorial Board Member of Journal and control algorithm of underwater gliders.
of Underwater Unmanned Systems and a member of
the Institute of IEEE Robotics and Automation. He was the recipient of the
National Excellent Youth Fund Winner, Young Scholar of the Yangtze River
of the Ministry of Education, and Tianjin Young and Middle-Aged Scientific
and Technological Innovation Leading Talent in China. In recent years, he has
taken charge of more than ten scientific projects, including the National Natural
Science Foundation of China, the National Key R&D Program of China, and the
National High-Tech R&D Program of China (863 Program). Toward underwater Xinhai Zhang received the B.S. degree from Yan-
equipment performance requirements of deep diving, long duration, and high shan University, Qinhuangdao, China, in 2018. He is
precision, he focuses his systematic research both on fundamental theories and currently working toward the M.S. degree with the
technical implementation. Some of his achievements, such as the underwater School of Mechanical Engineering, Tianjin Univer-
glider and Petrel, have been listed in China’s Top Ten Ocean Scientific and sity, Tianjin, China, both in mechanical engineering.
Technological Advances in 2015. He was also a recipient of the first prize of His research interest focuses on the effect of the
Tianjin Technological Invention Award in 2015, the second prize of the National biofouling on the glider performance.
Technical Inventions Award in 2016, and a Chinese Patent Award of Excellence.
Journals such as Popular Science and Armada International in the United States
have commented that Petrel is the first glider of China, and that Petrel creates
Chinese glider records of the voyage, depth, and endurance.

Authorized licensed use limited to: TIANJIN UNIVERSITY. Downloaded on March 01,2021 at 11:08:03 UTC from IEEE Xplore. Restrictions apply.
View publication stats

You might also like