Proceedings of ENCIT 2014 15th Brazilian Congress of Thermal Sciences and Engineering

Copyright © 2014 by ABCM November 10-13, 2014, Belém, PA, Brazil

TRANSIENT MODELING OF A SMALL HYDROKINETIC TURBINE

João José Albernaz Lopes, joaobernaz@gmail.com

Federal University of Pará , 01 Augusto Correa Street, Guamá, Belém, PA 66075-110, Brazil

Tércio Heitor de Sousa Moreira,moreira.tercio@yahoo.com.br

DjanirTravassosBrandão, djanirbrandao@hotmail.com

Sávio Weslley Oliveira Figueiredo, saviowes@hotmail.com

Jerson Rogério Pinheiro Vaz, jerson@ufpa.br

André Luiz Amarante Mesquita, andream@ufpa.br

Alexandre Luiz Amarante Mesquita, alexmesq@yahoo.com.br

Marcos André Barros Galhardo, galhardo@ufpa.br

Abstract. In recent years, increasing attention is being given to the study of hydrokinetic turbines for power generation
due to the use of clean energy by using renewable sources. This paper aims at to present a general methodology for the
efficient design of horizontal axis hydrokinetic turbines with variable rotation. The approach uses the Blade Element
Method (BEM) for determining the power coefficient of the turbine. The modeling of the hydrokinetic rotor is coupled
with the model of the drive line of the system, including the multiplier and the electric generator. Therefore, the
modeling of the whole system comprises the hydrodynamic information of the rotor and the characteristics of the
inertia of whole system, frictional losses and electromagnetic torque of the electrical generator. The results of
numerical simulation are obtained for the transient rotational speed of the rotor and compared with field data
surveyed from small hydrokinetic turbine installed at the Arapiranga-Açu creek.

Keywords: Hydrokinetic Turbine, Blade Element Method, Dynamic Modeling.

1. INTRODUCTION

This work develops a dynamic analysis of a small hydrokinetic turbine installed in the city of Acará, PA,which is
part of the microregion called Tome-Açu, located at latitude 01º57'39" South and longitude 48º11'48"West (see Fig.1a)
and at an altitude of 25 m. It has an area of 4363,6 km² and a population of 53605 habitants (IBGE, 2008).The
Arapiranga-Açu creek, shown in Fig.1b presents a velocity range of 0 to 1,2m/s, depending on the year season.
Readings of stream velocity and turbine shaft rotational speed for the cases with and without electrical loading
conditions were performed. After this step, it was done adaptation and application of the model developed by Mesquita
et al. (2013) to the dynamic analysis of hydrokinetic turbines.

Permanent magnet
alternator

Turbine

(a) (b)
Figure 1. (a) Location of the creek (b) Hydrokinetic turbine installed at Arapiranga-Açu.

The hydrokinetic turbine is an assembly of: (a) 4-blade rotor;(b) GÖ (Goetingen) 428/480 aluminum hydrofoil (480
on the tip and 428 on the base) of 60 cm diameter combined with a NACA 0012 airfoil; (c) Alternator (directly
connected to the pulley of the mechanical transmission system).Itsrated characteristics are: output power of 500 W,
rotation of 900 rpm and output voltage of 12 V, model A116, transmission ratio of 1:4, adopting efficiency of 96,5%,
according to Sarkis(1949). The greater pulley has a 400mm diameter, while the smaller has a 100 mm diameter, and the
clearance between shafts is 1265mm.
Proceedings of ENCIT 2014 15th Brazilian Congress of Thermal Sciences and Engineering
Copyright © 2014 by ABCM November 10-13, 2014, Belém, PA, Brazil

2. DYNAMIC MODEL

A hydrokinetic turbine system consists of a turbine rotor with mass moment of inertia JTconnected to a generator (load)
with mass moment of inertia JL through a gearbox with speed ratio r and efficiency η, as shown in Fig. 1.

Figure 1.Illustration of the complete system of a hydrokinetic turbine.

The shafts and the gears are considered to be infinitely rigid. Thisassumption is valid since the vibration modes of
the system are assumed to be in a frequency range higher than the operational frequency range. The dynamic equation
governing the power transmission system shown in Fig. is given by:

𝑑𝜔 𝑀
𝑇𝑀 = 𝑇𝑇 − 𝑇𝐷 − 𝑇𝐿−𝑀 = 𝐽𝑡𝑜𝑡𝑎𝑙 (1)
𝑑𝑡

The driving torque (TM) is equal to the torque of the turbine (TT) minus the dissipative torque (TD) (friction torques
of bearings). ωM is the angular speed of the rotor. The torque of the turbine is computed using hydrodynamic analysis.
TL-M is the equivalent load torque of the load. Its expression is given by:

1
TL-M = 𝑟𝑇𝐿 (2)
𝜂

where TL is the torque of the generator, η is the transmission efficiency and 𝑟 is the speed ratio 𝑟 = 𝜔𝐿 𝜔𝑀 . The total
mass moment of inertia of the system (Jtotal) is the sum of following moments of inertia: the inertia of the turbine (JT);
the inertia of the added mass of the fluid around theturbine blades (Jf), the inertia of the belt transmission with respect to
the input shaft ofthe multiplier (JGT), and the inertia of the generator (JL), i.e:

𝐽𝑡𝑜𝑡𝑎𝑙 = 𝐽𝑇 + 𝐽𝑓 + 𝐽𝐺𝑇 + 𝐽𝐿 (3)

In next subsections the other terms in Eq.(1) and (3) are discussed in details.Finally, the rotor torque T T can be
expressed as:

𝑃𝑀 1 𝜌𝜋 𝑅 2 𝑉 3
𝑇𝑇 = = 𝐶𝑃 (4)
𝜔𝑀 2 𝜔𝑀

where𝑃𝑀 is the output power, ρ is the density, 𝑅 is the radius of hydrokinetic turbine.

2.1. Generator Torque

Hong et al.(2013) and Hasanien(2010) show that the electricity generation systems are attracting great attention,
since those can be operated with constant speed or variable speed operations using power electronic converters. The
permanent magnet synchronous generator is a good option for the high performance in variable-speed generation
including high efficiency and high controllability for an electricity generation system. In this case, the torque of a
permanent magnet synchronous generator can be given through following algebraic equation:
3
𝑇𝐿 = 𝑝𝜓𝑖𝑠𝑞 (5)
2

where𝑝 is the pole pair number, 𝜓 is the magnet flux and 𝑖𝑠𝑞 is the electric current in one of the synchronous phases.

In order to reduce the complexity of the electromagnetic torque equation, it is assumed that the relation between
synchronous generator torque and its angular speed is given by an approximated linear equation (Bao and Ye (2001). In
Proceedings of ENCIT 2014 15th Brazilian Congress of Thermal Sciences and Engineering
Copyright © 2014 by ABCM November 10-13, 2014, Belém, PA, Brazil

this case, a first order linear function that describes the electromagnetic torque as a function of the generator angular
speed is used, which is given by:

𝑇𝐿 = 𝐾𝑒 𝜔𝑒 + 𝐾𝑒0 (6)

where𝐾𝑒 and 𝐾𝑒0 are obtained by a linear fit with experimental data.

2.2. Friction Torques

According to the loads acting on the bearings and the bearing types used throughout the system, it is possible to
estimate the overall friction torque in the system. Witte(1973) and Harris and Kotzalas(2006) empirically studied the
running friction torque of tapered roller bearings, which resulted in Eq.(10) for radial loaded bearings:
1
1
−11 𝐹 3
𝑇𝐷,𝑡𝑎𝑝𝑒𝑟𝑒𝑑 = 3,35 × 10 𝐺(𝜂𝑣0 )2 𝑓𝑡 𝑟 (7)
𝐾

where 𝐺 is a geometry factor based on the internal dimensions of the bearing, n is the rotational speed, 𝑣0 is the
lubricant kinematic viscosity, 𝑓𝑡 is a radial load factor and 𝐹𝑟 is the radial load. Palmgren(1959) and Harris and
Kotzalas(2006) separated the friction torque into a loading dependent component (𝑇1 ) and a load independent
component (𝑇0 ) which is influenced by the viscous property of lubricant type, the amount of the lubricant employed and
bearing speed. Thus, the total friction torque (in N.m) is given by Eq.(11):

𝑇𝐷,𝑟𝑜𝑙𝑙𝑖𝑚𝑔 = 𝑇0 + 𝑇1 (8)

where

𝑇1 = 10−3 𝑥𝑓1 𝑥𝐹𝛽 𝑥𝑑𝑀 (9)

2
𝑇0 = 10−10 𝑥𝑓0 𝑥(𝜂𝑣0 )3 𝑥𝑑𝑀
3
(10)

where 𝑓1 is a factor depending on the bearing design and relative bearing load, 𝐹𝛽 depends on the magnitude and
directions of the applied load, 𝑑𝑀 is the bearing pitch diameter and 𝑓0 is a factor depending on the type of bearing and
the method of lubrication. More detail about those formulas can be found in Harris and Kotzalas(2006). Ker Wilson
(1965) provides the following equation for the torque due to friction acting on all surfaces of a hollow disk:

𝜋𝜇 𝜔 𝑀 𝑅03 + 𝑅0 +2𝐿 −𝑅𝑖4

𝑇𝑣 = (11)
𝑕

where 𝑇𝑣 is the Frictional torque (in.lbf),𝜇 is the Fluid absolute viscosity (lbf.s/in2),𝜔𝑀 is the disk angular velocity
(rad/s), 𝑅0 is the disk outside radius (in), 𝑅𝑖 is the disk inside radius (in), h is the radial clearance or axial clearance
(assumed equal) (in) and L is the disk length (in).

2.3. Inertia Calculations

2.3.1.Turbine Rotor and Added Mass

The moment of inertia of the rotor of the turbine is given by:

𝐽𝑇 = 𝜂𝑏𝑙𝑎𝑑𝑒 + 𝐽𝑏𝑙𝑎𝑑𝑒 + 𝐽𝑕𝑢𝑏 (12)

where𝜂𝑏𝑙𝑎𝑑𝑒 is the number of blades, 𝐽𝑏𝑙𝑎𝑑𝑒 is the moment of inertia of a blade and 𝐽𝑕𝑢𝑏 is the moment of inertia of the
hub.
The moment of inertia of a blade is calculated as follows: the blade is divided into several small volumes along its
profile, and in each volume it is determined the center of gravity of the volume, the mass and the distance between the
center of mass to the center of rotation of the blade, as shown in Fig.3a. It is seen that in Fig.3a there is also the root of
the blade, which can be approximate as a thin cylindrical shell. Therefore, the equation that provides the moment of
inertia of the blade is given by:
𝑁 2 𝑁 2 2
𝐽𝑏𝑙𝑎𝑑𝑒 = 𝑖=1 𝑚𝑖 𝑟𝑖 + 𝐽𝑟𝑜𝑜𝑡 = 𝑖=1 𝑚𝑖 𝑟𝑖 + 𝑚𝑟𝑜𝑜𝑡 𝑟𝑟𝑜𝑜𝑡 (13)
Proceedings of ENCIT 2014 15th Brazilian Congress of Thermal Sciences and Engineering
Copyright © 2014 by ABCM November 10-13, 2014, Belém, PA, Brazil

The moment of inertia of the hub has a geometry shown in Fig.3, which can be approximated by a hollow
hemisphere. Thus, the expression of its moment of inertia is given by:

2 3
𝐽𝑕𝑢𝑏 = 𝑚𝑟𝑢𝑏 𝑟𝑟𝑢𝑏 (14)
3

For the mass of the cube,the masses corresponding to the holes that exist in the hub should be deducted.In the
expression of the total inertia of the system,the added mass of the fluid around the blades should also be taken into
account. In the model described by Maniaci and Li(2011), the added mass for a blade was assumed to be equal to the
mass of a cylinder (with length equal to the length of the blade 𝑅) whose diameter is equal to the chord length𝐶. Thus,
the expression of added mass is given by:

2 1
𝑚𝑎 = 𝜋𝑅𝑐𝑖𝑙 𝜌𝐿 = 𝜋𝐶 2 𝜌𝐿 (15)
4

where𝑅𝑐𝑖𝑙 is the radius of cylinder. In order to calculate the inertia of the fluid, it is proposed here the insertionof the
added mass calculated in Eq.(18) into Eq.(16) as follows

𝑁 𝑚𝑎 𝑚𝑎
𝐽𝑏𝑙𝑎𝑑𝑒 ,𝑓𝑙𝑢𝑖𝑑 = 𝑖=1 𝑚𝑖 + 𝑟𝑖2 + 𝐽𝑟𝑜𝑜𝑡 = 𝑁
𝑖=1 𝑚𝑖 + 𝑟𝑖2 + 𝑚𝑟𝑜𝑜𝑡 𝑟𝑟𝑜𝑜𝑡
2
(16)
𝑁 𝑁

Therefore, the moment of inertia of the rotor of the turbine and the added mass is given by:

𝐽𝑇 + 𝐽𝑓 = 𝜂𝑏𝑙𝑎𝑑 𝑒 𝐽𝑏𝑙𝑎𝑑𝑒 ,𝑓𝑙𝑢𝑖𝑑 + 𝐽𝑕𝑢𝑏 (17)

2.3.2. Generator Inertia

The equivalent inertia of the generator (JL−M) is given by:

1
𝐽𝐿−𝑀 = 𝑟 2 𝐽𝐿 (18)
𝜂

where𝐽𝐿 is the inertia of the generator, 𝜂 is the transmission efficiency and r is the speed ratio 𝑟 = 𝜔𝐿 𝜔𝑀 . For the
inertia of the generator rotor it was considered a cylinder of mass𝑚𝑔𝑒𝑛 , radius 𝑅𝑔𝑒𝑛 and length𝐿𝑔𝑒𝑛 , i.e:

2 1 4 1
𝐽𝐿 = 𝑚𝑔𝑒𝑛 𝑅𝑔𝑒𝑛 = 𝜌𝑔𝑒𝑛 𝜋𝑅𝑔𝑒𝑛 𝐿𝑔𝑒𝑛 (19)
2 2

3.RESULTS OF THE FIELD MEASUREMENTS

Figure 4 shows the measurements done in field. Ten measurements of rotation were done for each situation: rotation
with and without electrical loadingon the second shaft. For the measurement it was used a contact type tachometer. The
transmission ratio was 1:4, therefore the rotations on the turbine shaft are: 35rpm (with electrical loading) and 44,13
rpm(with no electrical loading). The water stream average velocity is 0,24m/s as shown in Fig.4b.
180 0.5
175

170 Average rotation = 176,51 rpm

0.4
Stream velocity- (m/s)

165
Rotation - (rpm)

Average velocity = 0,24 m/s

Without electrical load 0.3
160
With electrical load
155
0.2
150 Average rotation = 139,83 rpm

145 0.1
140 Measured vel.
Average vel.
135 0
0 2 4 6 8 10 0 2 4 6 8 10
Number of measurements Number of measurements
(a) (b)
Figure 4.(a) Alternator shaft rotations and (b) stream velocities measured at Arapiranga-Açu creek.

4. RESULTS OF THE DYNAMIC MODEL

Figure 5 shows values of rotation as function of time using the dynamic model previously depicted. It is observed
that in the region in which the function presents steady behavior, the result of the simulation is 1,75rad/s. In this case,
Proceedings of ENCIT 2014 15th Brazilian Congress of Thermal Sciences and Engineering
Copyright © 2014 by ABCM November 10-13, 2014, Belém, PA, Brazil

the relative error is 5% when compared to the mean values measured in field, shown in Fig.5. Such behavior is a proof
that the dynamic model has good concordance with the experimental results. For the unloaded case, it was not possible
to run the simulation by reason of numerical issues on the computational routine which is coupled with the dynamic
model, and is responsible for the determination of the turbine power coefficient. It is believed that the shaft rotational
speed and at low stream velocity, in this case 0,24m/s, causes problems in relation of the numerical model.

50
Present work
45
Measured rotatonal speed

Turbine rotational speed - (rpm)

40
Rotation (field measure) = 35 rpm

35
30
25 Rotation (numerical) = 33,24 rpm

20
15 relative error = 5%

10
5

0 10 20 30 40 50 60
Time - (s)
Figure 5. Rotation as a function of time.

5. CONCLUSION

In this paper, a comparison is done between the dynamic model proposed by Mesquita (2013) and the field data
measured at Arapiranga-Açu creek. The comparison reveals good concordance and a relative error of 5%. Such result
shows that the dynamic model described here is a useful tool for the efficient design of horizontal-axis hydrokinetic
turbines. However, it is noteworthy that more investigation on the numerical model and more measurements for greater
values of stream velocity are necessary in order to enlarge the range of results. The main limitations found in this work
are: (a) the numerical model presents failures on the low velocity condition and high rotations, and so needs more
attention; (b) the dissipative torque model is not applied to the machine start up velocity. It is believed that the
dissipative torque model needs modifications; (c) the data acquisition needs to be done for velocities close to 1 m/s in
order to complete the system power coefficient chart.

6. REFERENCES

Bao, N., Ye, Z., 2001.Active Pitch Control in Larger Scale Fixed Speed Horizontal Axis Wind Turbine Systems, Part I.
Linear Controller Design, Wind Engineering, Vol. 25, No. 6, pp. 339–999.
Betz, A.,1926.Wind-energy und ihreAusnutzungdurchWindmuehlen.
Budynas, Richard G; Nisbett, J. Keith, 2011.Elementos de Máquinas de Shigley – Projeto de
EngenhariaMecânica.8ª edition. Porto Alegre– McGraw-Hill.
Harris, T. A. and Kotzalas, M. N., 2006.Essential Concepts of Bearing Technology, 5 th edition, CRC Press.
Hasanien, H. M.,2010.Torque ripple minimization of permanent magnet synchronous motor using digital observer
controller. Energy Conversion and Management ; 51:98-104.
Hong, C. M.; Chen, C. H.; Tu, C. S.,2013. Maximum power point tracking-based control algorithm for PMSG wind
generation system without mechanical sensors. Energy Conversion and Management ,N° 69, pp 58-67.
Ker Wilson W., 1956. Practical Solution of Torsional Vibration problems, Volumes 1, 2, and 3, New York:John
Wiley and Sons, 3rd Ed.
Maniaci, D.C. and Li, Y., 2011.Investigating the Influence of the Added Mass Effect to Marine Hydrokinetic
Horizontal-Axis Turbines Using a General Dynamic Wake Wind Turbine Code, Oceans 11 Conference Kona, Hawaii
September, pp 19-21.
Mesquita, André L. A; Mesquita, Alexandre L. A; Vaz, Jerson R. P; Morais, Marcus. V. G. de, Gonçalves, Carmo,2013.
A Methodology for the Transient Behavior of Horizontal Axis Hydrokinetic Turbines.8th Conference on Sustainable
Development of Energy, Water, and Environment System, Dubrovnik, Croatia.
Palmgren, A., 1959. Ball and Roller Bearing Engineering, 3 rd ed., Burbank, Philadelphia, pp 34-41.
Sarkis; Melconian. Elementos de Máquinas, 7ª Ed.. São Paulo – Érica, 2000.
Witte, D., 1973. Operating Torque of Tapered Roller Bearings, ASLE Trans., Vol. 16,No 1, pp 61-67.

7. RESPONSIBILITY NOTICE
The following text, properly adapted to the number of authors, must be included in the last section of the paper:The
author(s) is (are) the only responsible for the printed material included in this paper.