www.usn.

no

FMH606 Master’s Thesis 2021

Electrical Power Engineering

Hydropower unit – An analysis of

operational efficiencies and energy losses

Sigurd S. Berg

Faculty of Technology, Natural sciences and Maritime Sciences

Campus Porsgrunn
 www.usn.no

Course: FMH606 Master’s Thesis 2021

Title: Hydropower unit – An analysis of operational efficiencies and energy losses
Pages: 148
Keywords: Weighted average efficiency, hydropower, efficiency, losses
Student: Sigurd S. Berg
Supervisor: Gunne J. Hegglid and Thomas Øyvang
External partner: Skagerak Kraft AS

Summary:
The beginning of the “green shift” has already started in the Nordic power grid. In this
context, it means extensive use of intermittent energy sources like wind and solar power.
The intermittent sources impact the operational regime of conventional hydropower units
(HPU) into unfavourable situations, resulting in reduced operational efficiency and
increased energy losses or revenue.
One of Norway’s largest power producers, Skagerak Kraft, has started a research
program collaborated with the University of South-Eastern Norway (USN) to investigate
the upcoming challenges related to HPUs and the “green shift”. This thesis will be the
beginning of this research program with the purpose of analysing Åbjøra and Sundsbarm
HPU concerning operational efficiency and investigate the efficiency under extended
operational regimes.
For the analyse, it has been made a static model in MATLAB that estimates and maps
the efficiencies and energy losses of the HPUs. The model determines the operational
efficiency based on the operational data from the generators. The data was provided by
Skagerak Kraft and contained information from the period 2020 with measurements of
active power, reactive power and voltage.
The results identified the best efficiency point (BEP) of 92.25 % and 92.14 % in Åbjøra
and Sundsbarm, respectfully, with an estimated weighted average efficiency (WAE) of 91.6
% for Åbjøra and 91.9 % for Sundsbarm. The analysis has shown a solid correlation between
reduced operational efficiency and frequent use of frequency restoration reserve (FRR),
also known as grid balancing. The magnitude of the active power has shown to be the
dominating factor related to efficiency, where operations at low active power attain the
lowest efficiencies. However, changes in reactive power could be assumed to have only
a minor effect on efficiency.

2
 1 Introduction

Preface
This master’s thesis concludes my Master of Science (MSc) degree within Electrical Power
Engineering, University of South-Eastern Norway (USN). This thesis covers a broad topic and
has given me an excellent introduction to the energy process of hydropower plants, and have
built a solid foundation for potential future work within the areas of hydropower.
I want to thank my supervisors Gunne J. Hegglid and Thomas Øyvang, for the support and
effort put into this thesis. I would also thank my fellow student from NTNU, Yannick C.
Karekezi, who has been very helpful with many great tips, in particular, regarding the generator
calculations.

Porsgrunn, 19th May 2021

Sigurd S. Berg

3
 1 Introduction

Contents
Preface ................................................................................................................... 3
Contents ................................................................................................................. 4
Nomenclature ...................................................................................................... 12
1 Introduction ..................................................................................................... 13
1.1 Background ........................................................................................................................... 13
1.2 Problem Statement ............................................................................................................... 13
1.3 Report structure .................................................................................................................... 14
2 Operation of the Nordic grid .......................................................................... 15
2.1 Operating regimes and energy production ........................................................................ 17
2.2 Complications to the future power grid.............................................................................. 18
2.2.1 Energy storage .............................................................................................................. 19
2.2.2 Intermittency .................................................................................................................. 20
2.2.3 Operational regime changes ........................................................................................ 21
2.3 The energy markets .............................................................................................................. 22
2.4 Production requirements and guidelines ........................................................................... 23
2.4.1 Functional requirements - Voltage limits ................................................................... 24
2.4.2 Functional requirements - Reactive capacity ............................................................. 25
2.4.3 Frequency requirements and energy reserves .......................................................... 25
3 System overview ............................................................................................. 27
3.1 Waterway ............................................................................................................................... 27
3.1.1 Construction .................................................................................................................. 27
3.1.2 Energy losses ................................................................................................................ 29
3.2 Hydropower turbine .............................................................................................................. 31
3.2.1 Main components – Francis and Kaplan turbines ..................................................... 31
3.2.2 Main components – Pelton turbines ........................................................................... 32
3.2.3 Energy losses – turbines.............................................................................................. 34
3.3 Generator ............................................................................................................................... 36
3.3.1 Construction of synchronous machines .................................................................... 36
3.3.2 Machine losses .............................................................................................................. 37
3.3.3 Capability diagram ........................................................................................................ 41
3.4 Transformer ........................................................................................................................... 42
3.4.1 Transformer losses ....................................................................................................... 42
4 Power plant description ................................................................................. 44
4.1.1 Åbjøra ............................................................................................................................. 45
4.1.2 Sundsbarm ..................................................................................................................... 47
5 Methods ........................................................................................................... 50
5.1 Simulation model .................................................................................................................. 51
5.2 Data acquisition and preparation ........................................................................................ 52
5.2.1 Data format .................................................................................................................... 52
5.2.2 Collecting and reading the CSV-files .......................................................................... 53
5.2.3 Data processing ............................................................................................................ 53
5.3 Waterway model.................................................................................................................... 54

4
 1 Introduction
5.4 Turbine model ....................................................................................................................... 55
5.5 Generator model ................................................................................................................... 57
5.5.1 Approximation of field current..................................................................................... 58
5.5.2 Capability diagram ........................................................................................................ 60
5.6 Transformer model ............................................................................................................... 63
5.7 Evaluation of hydropower generation ................................................................................ 64
5.7.1 Weighted average efficiency ........................................................................................ 64
5.7.2 Expected average efficiency ........................................................................................ 65
6 Power loss and efficiency analysis ............................................................... 67
6.1 Loss distribution ................................................................................................................... 68
6.2 Efficiency characteristics .................................................................................................... 70
6.3 Analysis of operational regime ........................................................................................... 76
7 Sensitivity analysis ......................................................................................... 84
7.1 Scenario 1 - Variation of active power ................................................................................ 85
7.2 Scenario 2 – Variation of reactive power ........................................................................... 89
8 Discussion ....................................................................................................... 91
8.1 Data acquisition and filtering process................................................................................ 91
8.2 Simulation model and assumptions ................................................................................... 91
8.2.1 Turbine and waterway model ....................................................................................... 92
8.2.2 Generator model ........................................................................................................... 92
8.2.3 Capability diagram ........................................................................................................ 93
8.3 Discussion of results............................................................................................................ 93
8.4 Discussion of sensitivity analysis ...................................................................................... 93
9 Conclusion ...................................................................................................... 95
10 Further work .................................................................................................... 96
References ........................................................................................................... 97

5
 1 Introduction

List of Figures
Figure 2.1: The Nordic power grid and the composition of energy sources. The size of the
circles corresponds to the magnitude of the energy production. Denmark (except eastern
Denmark) is not a part of the synchronous grid but connected through DC-links. Source:
NVE/[1].................................................................................................................................... 16
Figure 2.2: Weekly energy production and water inflow statistics of the Norwegian hydropower
[6]. ............................................................................................................................................ 18
Figure 2.3: Estimated growth in wind and solar power in the Nordic [7]. .............................. 19
Figure 2.4: The “Duck curve”, a typical representation of the net load variation in power grids
with a high solar generation [8]. .............................................................................................. 19
Figure 2.5: Wind power production profile for a single turbine and an entire wind park over
one month [9]. The bottom figure shows production data for a find farm, whereas the upper
figure shows a single turbine from the same farm and under the same period. ....................... 20
Figure 2.6: Solar power production profile over one day [10]. ............................................... 21
Figure 2.7: Distribution and density of load points of 300 MVA generator with highly volatile
utilization, illustrating the situation before (a) and after (b) the “German Energiewende” [11]
[2]. ............................................................................................................................................ 22
Figure 2.8: System price calculation, where Ps is the spot price. ............................................ 23
Figure 2.9: The response time of the different reserves [19]. .................................................. 26
Figure 3.1: Cleaning process of a sand trap in Rosekrepp hydropower [21]. .......................... 28
Figure 3.2: A multiple penstock arrangement [22]. ................................................................. 28
Figure 3.3: Layout of an HPU, illustrating the relationship between the gross head, net head
and head loss [24]. ................................................................................................................... 29
Figure 3.4: Illustration of local losses produced in a pipe with sudden contraction and the rise
of additional turbulence [27]. “Vena contracta” is referred to as the point where fluid velocity
is at its maximum. .................................................................................................................... 30
Figure 3.5: The cross-section and the components of a) Francis turbine and b) Kaplan turbine
[29] [30]. .................................................................................................................................. 32
Figure 3.6: Cross-section of a Pelton bucket where the water jet reflects water at an angle of
165°. The figure is based on a figure given in [31]. ................................................................ 33
Figure 3.7: Construction of a five nozzle Pelton turbine [32]. ................................................ 33
Figure 3.8: Helical vortices formation in a draft tube at (a) part-load and (b) full-load regime
[34]. .......................................................................................................................................... 35
Figure 3.9: Energy loss in the form of leakage [35]. ............................................................... 35
Figure 3.10: Illustration of a salient-pole design of a large synchronous generator, used in an
HPU [38] .................................................................................................................................. 36

6
 1 Introduction
Figure 3.11: Design of (a) cylindrical rotor and (b) salient pole rotor [39]. ............................ 37
Figure 3.12: Armature core loss relative to the applied terminal voltage, as described in the
IEEE STD 115 [42]. ................................................................................................................. 39
Figure 3.13: Illustration of a hysteresis loop [46].................................................................... 39
Figure 3.14: Armature winding loss and stray-load loss, as described in the IEEE STD 115 [42]
.................................................................................................................................................. 40
Figure 3.15: Illustration of a capability diagram to a salient pole generator. .......................... 42
Figure 3.16: Design of a large power transformer (ABB’s TrafoStar design) [51]. ................ 43
Figure 4.1: Disassembly of the old Pelton turbines in Åbjøra (2007). [Photo: Jan Erik Olsrud]
.................................................................................................................................................. 45
Figure 4.2: Inside of Åbjøra power station, showing the generator. [Photo: Skagerak Kraft]46
Figure 4.3: Simplified overview of the waterway in Åbjøra, where the numbers (1-9) represents
the different elements used in the measurements found in Table 4.1. In the figure one can see
the conduit and penstock is divided into two whereas the tailrace is divided into three elements.
.................................................................................................................................................. 46
Figure 4.4: Inside of Sundsbarm power station, showing the generator. [Photo: Skagerak Kraft]
.................................................................................................................................................. 48
Figure 4.5: Simplified overview of the waterway in Sundsbarm. The element numbers (1-4)
represents the object element found in Table 4.3. ................................................................... 48
Figure 5.1: Diagram describing the structure of the simulation model. .................................. 50
Figure 5.2: Structure of the HPU model. ................................................................................. 51
Figure 5.3: Power flow diagram, illustrating the calculation direction, where Pi represents the
electrical production (input data) of the transformer and Po represents the hydraulic power
entering the waterway (output data from the model). .............................................................. 51
Figure 5.4: An illustration of the efficiency curve of a Francis turbine, where the dotted line is
the given data points, and the red line is the interpolated curve. ............................................. 56
Figure 5.5: Efficiency mapping of Sundsbarm, with bilinear interpolation between active power
(𝑃), gross head (𝐻𝑔𝑟𝑜𝑠𝑠) and total plant efficiency (𝜂𝐻𝑃𝑈). ................................................. 56
Figure 5.6: Armature current limit and maximum active power (𝑃𝑚𝑎𝑥) where 𝑃𝑚𝑎𝑥 = 𝑆𝑛 ∙
cos𝜑𝑛.The figure is based on [49]. .......................................................................................... 60
Figure 5.7: Field current limit. The figure is based on [49]. .................................................... 61
Figure 5.8: Capability diagram including practical stability limit, theoretical stability limit, end
region heating limit (k=0.3), field current limit (k=1) and armature current limit. The dotted
curves represent the field current under different excitation voltages (k=0.1-1). The figure is
based on [49] ............................................................................................................................ 63
Figure 5.9: The time-depending efficiency trend for WAE calculation [8]. ........................... 65

7
 1 Introduction
Figure 5.10: Illustration of EAE, where EAE is the WAE is estimated for all operations within
the green area, here 91 %. [own work] .................................................................................... 66
Figure 6.1: Chapter structure. .................................................................................................. 67
Figure 6.2: Efficiency curves in relation to power output (measured at transformer output) in
Åbjøra. Calculated with: 𝑐𝑜𝑠𝜑𝑖𝑛𝑑 = 0.9, 𝑉 = 11 𝑘𝑉 and 𝐺𝑟𝑜𝑠𝑠 ℎ𝑒𝑎𝑑 = 442𝑚. ............... 68
Figure 6.3: Loss distribution under nominal generator operating point of a) Åbjøra and b)
Sundsbarm................................................................................................................................ 69
Figure 6.4: Power loss in the waterway (Åbjøra), showing the relationship between power loss
[𝑀𝑊] and flow rate [𝑚3/𝑠] relative to gross head [𝑚] under constant power (95 MW). ...... 70
Figure 6.5: Iso-efficiency contour map of the Francis turbine used in Åbjøra. ....................... 71
Figure 6.6: Iso-efficiency contour map of the entire HPU, showing the relationship between
efficiency, active power and head in a) Åbjøra and b) Sundsbarm. Reactive power and terminal
voltage are set to nominal values. ............................................................................................ 72
Figure 6.7: Efficiency mapping of the synchronous generator used in Åbjøra. ...................... 73
Figure 6.8: Iso-efficiency contour map of the entire HPU, showing the relationship between
efficiency, active and reactive power under the optimal gross head, in a) Åbjøra and b)
Sundsbarm................................................................................................................................ 75
Figure 6.9: Production regime of Åbjøra, showing both active and reactive production. ....... 76
Figure 6.10: Production regime of Sundsbarm, showing both active and reactive production.
.................................................................................................................................................. 77
Figure 6.11: Distribution of operational data for a) Åbjøra and b) Sundsbarm. The dotted
horizontal line in the left figure represents the generator's nominal rating (active power). The
figure to the right shows the rough distribution of all operating points................................... 78
Figure 6.12: Cumulative probability of all operational efficiencies for both Åbjøra (blue) and
Sundsbarm (red). The EAE is marked as large, dotted lines. Operations situated in off-state are
not considered in the figure...................................................................................................... 79
Figure 6.13: A hypothetical scenario of Åbjøra showing the effect of shifting the turbine
characteristic curve. The turbine is given by interpolated measurements, which have been
multiplied by 0.95, 1.0 (original) and 1.05 in order to move the turbine characteristic curve.
The intersection point between HPU characteristic (blue) and average production is marked
with a red circle. ....................................................................................................................... 80
Figure 6.14: WAE of Åbjøra and Sundsbarm. ......................................................................... 80
Figure 6.15: Production regime Åbjøra during January and April, with their representative
WAE compared to the BEP. The hours are referenced from the first measurement (22.01.2020).
.................................................................................................................................................. 81
Figure 6.16: Production regime Sundsbarm during August and July, with their representative
WAE compared to the BEP. The hours are referenced from the first measurement (22.01.2020).
.................................................................................................................................................. 81

8
 1 Introduction
Figure 6.17: The monthly energy production and energy loss in a) Åbjøra and b) Sundsbarm.
.................................................................................................................................................. 82
Figure 6.18: Energy loss a) Åbjøra and b) Sundsbarm. The optimal energy loss represents the
energy loss that would be obtained if operated at BEP under equal production as the actual
data. .......................................................................................................................................... 83
Figure 7.1: Illustration of the sensitivity analysis scenarios. Green area represents the original
operational regime, whereas red and yellow area represents expansion and contraction of the
operational regime, respectfully............................................................................................... 84
Figure 7.2: Scenario 1, variation of active power, where left figures represent operations of
Åbjøra and right represents Sundsbarm. Test P1 shows full contraction (𝑎𝑖 = 0.2) and test P10
shows full extension (𝑎𝑖 = 2). ................................................................................................. 86
Figure 7.3: Estimated WAE for each test ................................................................................ 87
Figure 7.4: Correlation between average active power, BEP and WAE obtained from each test
for a) Åbjøra and b) Sundsbarm............................................................................................... 88
Figure 7.5: Increased variation of reactive power, where a) is the operation in Åbjøra and b) is
Sundsbarm. Test Q11 represents a variation of reactive power three times the original......... 89
Figure 7.6: WAE obtained from each test for both Åbjøra and Sundsbarm. The test nr.
represents each simulation where all reactive power measurements are multiplied by a constant
ranging from 1.0 – 3.0.............................................................................................................. 90
Figure 10.1: Moody diagram. A diagram used to determine friction factor (f) from the relative
roughness factor and the Reynolds number [58]. .................................................................. 106
Figure 10.2: Illustration of head losses shown in sections..................................................... 108

9
 1 Introduction

List of Tables
Table 2.1: Limits for thresholds for type A – D power-generating modules in the Nordic area
[12] [13]. .................................................................................................................................. 24
Table 2.2: Requirement for the minimum operative period for synchronous power-generating
modules related to voltage limits [12] [13]. ............................................................................. 24
Table 2.3: Reference voltage related to the Norwegian grid voltage [13]. .............................. 24
Table 2.4: General requirement to reactive capacity for synchronous power-generating
modules, referred to PCC [13]. ................................................................................................ 25
Table 2.5: General requirement to reactive capacity for synchronous power-generating
modules, referred to generator terminals, assuming 12 % reactive consumption in transformer
[13]. .......................................................................................................................................... 25
Table 2.6: Requirements for the minimum operative period for synchronous power-generating
modules, related to frequency limits under the voltage range 0.9 – 1.05 Pu [13]. .................. 25
Table 3.1: Typical specifications of turbines ........................................................................... 31
Table 4.1: Overview of the nominal specifications to Åbjøra and Sundsbarm hydropower
station. * Mechanical losses in Sundsbarm contains windage, ventilation and bearing losses.
.................................................................................................................................................. 44
Table 4.2 Head loss measurements of each element in the waterway from Åbjøra ................ 47
Table 4.3: Head loss measurements of each element in the waterway from Sundsbarm. ....... 47
Table 5.1: Example of the filtering process of the voltage to Sundsbarm. The “normal” value
or the maximum trigger voltage is here chosen to be 14.4 kV. In operation nr. 2, the
𝑑𝑢𝑟𝑎𝑡𝑖𝑜𝑛 = 9.59/14.4 = 0.67 or rounded to 3/4, which can be seen in step 4. ................... 54
Table 6.1: Summary of operational statistics of Åbjøra and Sundsbarm. Operating time
determined from nominal turbine power and total energy production. Results from the period
(22.01.20 – 31.12.20). .............................................................................................................. 67
Table 6.2: Optimum operation points at Åbjøra ...................................................................... 75
Table 6.3: Optimum operation points at Sundsbarm ............................................................... 75
Table 7.1: Overview of test values used .................................................................................. 85

10
 1 Introduction

Abbreviations
AVR Automatic Voltage Regulator
BEP Best Efficiency Point
CFD Computational Fluid Dynamic
CSV Comma-Separated Value
EAE Expected Average Efficiency
ENTSO-E European Network of Transmission System Operators for Electricity
FCR Frequency Containment Reserve
FEM Finite Element Method
FFR Fast Frequency Reserve
FIKS Funksjonskrav i kraftsystemet (Functional requirements in the power system)
FRR Frequency Restoration Reserve
HPU Hydropower Unit
IEC International Electrotechnical Commission
IEEE Institute of Electrical and Electronics Engineers
NC-RfG Network Code on Requirements for Generators
NVE Norges vassdrags- og energridirektorat (The Norwegian Water Resources and
Energy Directorate)
NVF Nasjonal veileder for funksjonskrav i kraftsystemet (National guide for
functional requirements in the power system)
OCC Open Circuit Characteristic
SCADA Supervisory Control and Data Acquisition
TSO Transmission System Operator
WAE Weighted Average Efficiency

11
 1 Introduction

Nomenclature
Symbol Description
𝑉 Voltage [𝑘𝑉] or [𝑃𝑢]
𝐼 Current [𝐴] or [𝑃𝑢]
𝑅 Resistance [Ω] or [𝑃𝑢]
𝑃 Active power [MW] or [Pu]
𝑄 Reactive power [MVAr] or [Pu]
𝑆 Apparent power [MVA] or [Pu]
𝜑 Power factor angle [𝑅𝑎𝑑] or [𝐷𝑒𝑔]
𝑃𝐹 Power Factor
𝑄𝑓𝑙𝑜𝑤 Volumetric flow rate [𝑚3 /𝑠 ]
𝐻 Head [𝑚]
𝑣 Velocity [𝑚/𝑠]
𝐾 Total head loss coefficient
𝑘𝑖 Head loss coefficient of a section
𝐸, 𝐸𝑝 Excitation voltage, Potier voltage [𝑘𝑉] or [𝑃𝑢]
𝑋𝑑 , 𝑋𝑞 , 𝑋𝑑′ , 𝑋𝑙 , 𝑋𝑝 Reactance d-axis, q-axis, transient, leakage and Potier [Pu]
𝛿 Load angle [𝑅𝑎𝑑] or [𝐷𝑒𝑔]
𝜂 Efficiency [%]
𝑇𝑠 , 𝑇0 Temperature [℃]
𝛼𝑡 Temperature coefficient of resistance [(℃)−1]
𝑏𝑣 The slope of the air-gap line [𝑃𝑢/𝐴] or [𝑃𝑢/𝑃𝑢]
𝐴𝑘 Weighting factor
𝑓 Friction factor
𝐷 Diameter [𝑚]
𝑔 Gravitational constant [𝑚/𝑠 2 ]
𝐿 Length [𝑚]
𝜌 Water density [𝑘𝑔/𝑚3 ]
𝑇 Total duration time [ℎ]
∆𝑡𝑘 Duration of a loading point [ℎ]

12
 1 Introduction

1 Introduction
The Norwegian power grid is mainly powered by hydropower with a share of about 90 %,
where 75 % of all hydropower plants are so-called impoundment or reservoir plants[1]. The
operating regimes of impoundment types are usually marked based, i.e., their production is
carefully scheduled to the energy market called Nord Pool to optimise profit. Some hydropower
plants may also be in the so-called capacity market, a market used to balance and stabilise the
grid. Hydropower has a unique feature, particularly impoundment power plants can store a
significant amount of energy that can be used whenever there is a need. Thus, with the large
share of hydropower, Norway or the Nordic grid can utilise energy sources with high
intermittencies, such as wind and solar power plants. Although hydropower can handle
excessive use of wind and solar power, it can negatively influence the operational regime and
affect efficiency, which will lead to reduced revenue.

1.1 Background
In recent years, several situations have been identified where conventional power plants have
had their operating regimes changed due to extensive use of renewables like solar and wind
[2]. This ongoing shift exposes new operating regimes with many start/stop cycles, extended
range of operation and large voltage fluctuations [2]. Earlier studies [3] have analysed the effect
of greater wind power penetration in transmission constrained areas. The study showed a
possible reduced revenue to hydropower producers, particularly impoundment power plants,
which are forced to operate with higher variations in production. A hydropower plant consists
of turbines and generators sensitive to changes in the operational regime. These are optimised
to a specific operating point, usually around the rated power. Deviating far from this point may
cause a steep reduction of efficiency and large energy losses, energy which could have been
stored and utilised in periods with a higher market price.

Skagerak Kraft has shown interest in this topic by starting a research program collaborated
with the University of South-Eastern Norway (USN) to investigate losses and efficiency in
their hydropower plants to gain better insight into possible future losses.

1.2 Problem Statement

The main goal of this project is to analyse the operational efficiency and energy losses of
specific hydropower units (HPU) based on today's and future operating regime. The objectives
of this thesis are listed below. The task description for this thesis was formally adjusted, where
the first iteration can be found in Appendix A. The corrections are covered in the minutes of
meeting to the formal meeting, found in Appendix B.

1. Review the existing regulation and requirements and typical operating conditions of the
Norwegian Bulk Hydropower.
2. Explore the losses (energy conversion) of a hydropower system.

13
 1 Introduction
3. Build a suitable mathematical model of the HPU (waterway, turbine, generator and
transformer) to identify energy loss and efficiency under stationary conditions.
4. Identify the optimal operating point/pattern of the HPU regarding efficiency. This from
operation statistic for specific units for one or some years.
5. Run different cases based on future predictions to identify energy loss under new
operating regimes.

1.3 Report structure

The report is structured to provide the reader with a general understanding of loss, regulation,
construction and the operation of a hydropower units, followed by an in-depth analysis of
specific hydropower units with respect to losses and their operation. A summary of each
chapter’s content is as follows:

Chapter 2 will give a short overview of the operating regimes and energy production in today’s
hydropower units and the future challenges regarding wind and solar. In addition, a brief
description of the energy market and the regulations relevant for hydropower units, concerning
limits in energy production.

Chapter 3 will give the reader a basic understanding of the main components of a hydropower
unit concerning function, construction and energy losses.

Chapter 4 will present general background information about the hydropower plants analysed
and their specifications.

Chapter 5 will present a description of the methods used in this analysis.

Chapter 6 will present an in-depth analysis of the specific hydropower units with a focus on
operational efficiencies and energy losses.

Chapter 7 will present a brief sensitivity analysis under selected operational scenarios to
identify changes to the average efficiency in possible future operating regimes.

Chapter 8 will present the discussion of the methods used and for the simulation results
presented in chapter 6 and 7.

Chapter 9 will present the conclusions.

Chapter 10 will present the future work.

14
 2 Operation of the Nordic grid

2 Operation of the Nordic grid

This chapter gives an overview of today's and possible future operating regime of HPUs and
an introduction to the energy market and national and international regulations, focusing
primarily on synchronous generators and their operation in the Norwegian grid. The energy
market and the regulations are vital factors that describe the operating regimes of hydropower
plants and are therefore described in this thesis.
The inter-Nordic system collaborates between Norway, Sweden, Finland and eastern Denmark,
creating one large synchronous power grid, depicted in Figure 2.1. The energy composition in
the Nordic gird consists of mostly hydropower, nuclear, wind and thermal, where today’s
shares of hydropower production in Norway, Sweden and Finland consist of around 90 %, 40
% and 23 %, respectfully [1] [4]. The Nordic grid utilises HPUs with energy storage primarily,
often referred to as impoundment plants. The impoundment plants are, in particular, essential
for the Norwegian power grid, which allows sufficient production throughout the year.
However, the HPUs are often used to balance the grid. The balance might heavily be affected
by what is known as the “green shift”, where intermittent energy sources, like wind and solar
power, become excessively used and can cause a significant imbalance.
Although the Nordic grid, notably Norway, has large energy storage in HPU, it must be enough
stored energy in periods with a limited inflow. Thus, the energy market is valuable. It maintains
a balance between supply and demand and ensures cheap run-of-river plants before the energy
storages essential for periods with limited inflow, like in the winter season. In addition, there
are national and international standards established for having an interconnected synchronous
grid collaborated between multiple nations and numerous generators, which will also affect the
operational regimes of the power plants. The standards and requirements discussed in this
chapter will focus primarily on synchronous generators and their operation in the Norwegian
grid.

15
 2 Operation of the Nordic grid

Figure 2.1: The Nordic power grid and the composition of energy sources. The size of the circles corresponds to
the magnitude of the energy production. Denmark (except eastern Denmark) is not a part of the synchronous
grid but connected through DC-links. Source: NVE/[5].

16
 2 Operation of the Nordic grid
2.1 Operating regimes and energy production
There are two main types of hydropower units, regulated and unregulated, where in Norway,
the regulated types account for about 75 % of the total hydropower production [1]. The most
common type of regulated and unregulated power plants in the Nordic countries are the
impoundment (reservoir) type and the run-or-river type.

Impoundments plants store water in natural or artificial lakes at high elevations and often seen
with dams to increase the reservoir capacity. This feature allows the power plants to produce
power whenever there is a need, assuming there is enough water in the reservoir. With the
exceptional controllability of impoundment plants, production scheduling becomes possible,
allowing the power plants to plan or predict their operation into the future to optimise profit.

Run-of-river plants are often located in or close to rivers, sometimes in combination with dams
to increase the headwater allowing for a higher level of control to the production. The downside
to this type of plants is their limited controllability, i.e., the inflow of water has to be equal or
close to identical to the water discharged through the turbines at all times. Otherwise, water
will overflow and get wasted.

In the context of energy production, one may divide the term power plants into:
1. Baseload power plants.
2. Load-following power plants.
3. Peak power plants.
Baseload power plants operate with constant or maximum power output and are often
associated with run-of-river plants. The operation of a run-of-river plant is usually equal to the
inflow as there are often little to no storage capacity and has an extended operational range,
resulting in a low utilisation time of around 3000-4000 hours. Impoundment plants with a high
mean annual inflow can also be considered base load power plants or load-following power
plants. These impoundment plants usually operate with a close to constant power output,
resulting in a high utilisation time of around 6000-7000 hours. In Norway, these type of power
plants was generally used in combination with energy-dense industries like steelworks which
had a high baseload with minor variations in consumption.

Peak power plants are, in comparison, power plants used to operate under high demand mainly.
Some impoundment plants are considered peak power plants, with a large capacity to utilise
the annual inflow within a short utilisation time, often in the range of 1000-2000 hours. The
Norwegian power grid uses hydropower primarily with a quick response time and large energy
reserves, which has led to limited interests in peak power plants in Norway. Peak power plants
usually have high investment costs, mainly due to the increased unit cost, and are therefore
located where the head is high, head losses are low, and the ability for regulation is good.
The utilisation times described in the last two paragraph’s was stated by the main supervisor
Prof. Hegglid, by mail (27.04.2021).

17
 2 Operation of the Nordic grid

The total energy production and water inflow of all hydropower units in Norway are shown in
Figure 2.2. How reservoirs utilise and distribute the stored energy throughout the year is
illustrated by the figure. When the water inflow (measured as energy) is greater than the energy
production or consumption, the water fills the reservoirs. The filling process occurs during the
late spring to late autumn due to severe rainfall and snow melting from the mountains.

Figure 2.2: Weekly energy production and water inflow statistics of the Norwegian hydropower [6].

2.2 Complications to the future power grid

In today’s “green shift”, one can expect a radical transformation of the power grid in the future
power grid due to the extensive use of renewable energy sources and other technologies. The
change may influence the operating regimes of today's conventional hydropower units,
particularly impoundment plants, which are often used to balances the grid (FRR). The energy
balance and stability of the grid might be limited and reduced with excessive use of solar and
wind power plants as these are considered to have ha high intermittency, i.e., irregular and
unpredictable production.
A long-term power market analysis given by NVE [7] speculates that there will be a significant
increase in wind and solar power for the Nordic nations, as shown in Figure 2.3. The analysis
estimates wind and solar power shares in the Nordic region would increase from around 20 %
in 2020 to approximately 40 % in 2040 [7].

18
 2 Operation of the Nordic grid

Figure 2.3: Estimated growth in wind and solar power in the Nordic [7].

2.2.1 Energy storage

Countries with excessive use of solar and wind production have difficulties regarding energy
storage in the grid. The so-called “Duck curve”, as depicted in Figure 2.4, describes the net
load variation, i.e., the imbalance between peak demand and renewable (solar) energy
production. The owner of such a grid may invest in additional peak power plants to support the
high demand when renewable sources are inactive. Thus, hydropower, particularly
impoundment plants, would be ideal to be combined with solar power to balance the
production.

Figure 2.4: The “Duck curve”, a typical representation of the net load variation in power grids with a high solar
generation [8].

19
 2 Operation of the Nordic grid

Excessive use of wind power may also result in challenges regarding energy storage as the
wind does not always blow enough to run the wind turbines. Fortunately for Norway and
Sweden, the power grid has a large energy storage capacity. The master thesis of Bødal [3]
analysed how the coordination of hydro and wind power in a transmission constrained area
would affect the revenue to the power producers. The thesis concludes that hydro and wind
power coordination would result in lower power levels for impoundment plants as production
moves to benefit run-of-river plants. The increased penetration of wind might give a small
reduction in the total profit for impoundment hydropower, whereas the run-of-river would
expect increased profit [3].

2.2.2 Intermittency
Another challenge associated with the excessive use of wind and solar is their intermittency.
Intermittency is a common term used to describe the irregular and unpredictable production of
wind and solar. There has been plenty of research on predicting both wind and solar production
from weather data in recent time. Prediction of the production could help planning ahead of
production coordination, but the irregular output of wind and solar will still be present, as
depicted in Figure 2.5 and Figure 2.6. In both figures, one can see large intermittency as the
production is highly irregular and unpredictable. In Figure 2.5, one can see that an entire wind
farm will have more or less the same intermittency as a single turbine. As shown in Figure 2.6,
the solar power production illustrates a predictable pattern that follows the sun but has irregular
production, most likely due to clouds. Wind or solar farms with high intermittency can result
in what is generally referred to as a “voltage dip” and is one of the main concerns for every
TSO [9]. In most cases, the low voltage or dips are resolved by the reactive power controllers,
thus increasing the fluctuations in the reactive production regime for the conventional power
plants. In the worst-case scenarios, the voltage dip may lead to island operation or blackouts.

Figure 2.5: Wind power production profile for a single turbine and an entire wind park over one month [9]. The
bottom figure shows production data for a find farm, whereas the upper figure shows a single turbine from the
same farm and under the same period.
20
 2 Operation of the Nordic grid

Figure 2.6: Solar power production profile over one day [10].

2.2.3 Operational regime changes

A synchronous generator situated in a conventional power plant could be subjected to
significant effects of increased wind and solar penetration, as illustrated in Figure 2.7. The
figure is an illustration that shows the before and after the so-called “German Energiewende”,
where one can see significant changes to the active and reactive power [11] [2]. Still, most
importantly, the distribution and changes in operational density are considerably affected. The
difference in the operating regime as illustrated in Figure 2.7 can be considered an extreme
scenario, especially for most conventional hydropower plants. In HPUs, the turbine usually
limits the maximum and minimum active power production and will most likely not change
significantly in the future. On the other hand, the future operating regime for hydropower plants
may experience more significant fluctuations in production, more start/stop situations, and
different operational density distribution. These changes may go at the expenses of reduced
operating efficiency and more fatigue in several components.

21
 2 Operation of the Nordic grid

Figure 2.7: Distribution and density of load points of 300 MVA generator with highly volatile utilization,
illustrating the situation before (a) and after (b) the “German Energiewende” [11] [2].

2.3 The energy markets

Nord Pool, the Nordic energy market, exchanges electric energy between the Nordic, Baltic,
Central Western Europe and the UK. The function of Nord Pool is to stimulate the production
and the demand to maintain a balanced grid. Nord Pool consists of three physical markets:
- day-ahead (Elspot)
- intraday (Elbas)
- balance services
where most of the energy sold is through the Elspot-market. In the Elspot-market, the power
producers offer a given amount of power they would deliver the next day. At the same time,
large industries and energy companies assess the consumption, which is likely to occur the next
day. An algorithm then estimates the price based on the intersection between the aggregated
supply and demand curves, as shown in Figure 2.8.

22
 2 Operation of the Nordic grid

Figure 2.8: System price calculation, where Ps is the spot price.

The demand estimated the day ahead might not reflect the actual consumption and produce
some errors balanced through adjustments in the Elbas-market. However, the price will not be
equal for all countries or even within a nation as the price shall stimulate and balance the supply
and demand. A typical situation is a power line connecting two areas that exceed its capacity,
also known as a bottleneck. Thus, the two areas shall have a price differently from one another
to retain balance by reducing the price in areas with surplus and decrease the deficit area's cost.
The last markets are the balancing services where power participants get paid either from
delivered energy from automatic grid balancing (FFR, FCR or aFRR) or by manual grid
balancing mFRR, also known as “regulerkraftmarkedet” (RK), see section 2.4.3. In addition,
power participants can also be in the so-called capacity market or
“regulerkraftopsjonsmarkedet” (RKOM), where they get paid for available capacity, often
seasonal contracts.

2.4 Production requirements and guidelines

This section shall give an overview of the requirements and guidelines that briefly describe the
operating regimes for production units, particularly hydropower units. All generators
connected to the power grid is obligated to follow specific requirements. In Europe and the
Nordic power grid, the ENTSO-E (European Network of Transmission System Operators for
Electricity) is the authorized entity. For power producers, there is an under category in ENTSO-
E called NC-RfG (Network Code on Requirements for Generators) [12], which are connection
rules for power-generating modules1. In Norway, the NVE (The Norwegian Water Resources
and Energy Directorate) is the national authorized entity based on the ENTSO-E with
modifications customized to the Norwegian power grid. All power-generating modules
connected to the Norwegian power grid shall follow the technical requirements provided by
the national TSO (Statnett). Statnett guides producers to follow the regulations through
guidelines, in particular, the new guideline NVF (National guide for functional requirements

1
An installation which generates electricity, e.g., a hydropower plant.

23
 2 Operation of the Nordic grid
in the power system) [13], which took over for the old guideline FIKS (Functional requirements
in the power system) [14] in late 2020. The requirements for generator design and
calculation/measuring methods are given by the IEEE standard C50.13 [15] and IEC
(International Electrotechnical Commission) standard 60034 [16].
In the NC-RfG, the maximum capacity of an installation and its maximum voltage at PCC
(point of common coupling) will determine which requirements to follow. The requirements in
NC-RfG is structured so that type A power-generating modules have the least requirements,
and B, C and D will have an additional set of rules in the respectful order. The power-generating
modules for the Nordic area are categorised as shown in Table 2.1, which are also the origin of
the models described in the national guideline NVF.
Table 2.1: Limits for thresholds for type A – D power-generating modules in the Nordic area [12] [13].
Type Voltage level at Maximum capacity
PCC
A < 110 kV 0.8 kW
B < 110 kV 1.5 MW
C < 110 kV 10 MW
< 110 kV 30 MW
D
≥ 100 kV All installed capacities

2.4.1 Functional requirements - Voltage limits

NVF and NC-RfG have specified that all synchronous power-generating modules connected to
a power grid shall be able to operate within the voltage range described in Table 2.2. The table
shows the minimum periods during which a power-generating module must be capable of
operating for voltages deviating from the reference 1 Pu value at the connection point without
disconnecting from the network [12]. For the Norwegian power grid, the NVF has specified
the continuous voltage boundaries for voltage levels of 66 kV and 420 kV, related to the
reference voltage as seen in Table 2.3.
Table 2.2: Requirement for the minimum operative period for synchronous power-generating modules related to
voltage limits [12] [13].
Synchronous area Voltage range Time period for operation
0.90 Pu – 1.05 Pu Unlimited
Nordic
1.05 Pu – 1.10 Pu 60 minutes
Table 2.3: Reference voltage related to the Norwegian grid voltage [13].

Nominal 0.9 Pu minimum 1.0 Pu 1.05 Pu Maximum

system voltage continuous voltage Reference voltage continuous system voltage
420 kV 360 kV 400 kV 420 kV
300 kV 256 kV 285 kV 300 kV
132 kV 125 kV 138 kV 145 kV
110 kV 105 kV 117 kV 123 kV
66 kV 62 kV 69 kV 72.5 kV

24
 2 Operation of the Nordic grid
2.4.2 Functional requirements - Reactive capacity
Synchronous power-generating modules must have a minimum reactive capacity to ensure high
voltage stability. The minimum requirement for reactive capacity is given by the maximum
power, 𝑃𝑚𝑎𝑥 at nominal voltage, 𝑈𝑃𝐶𝐶 = 1.0 𝑃𝑢, referred at PCC [13], see Table 2.4. The
reactive power shall not be limited when 𝑃 < 𝑃𝑚𝑎𝑥 [13]. In practice, one could assume a
reactive consumption of about 12 % in for the transformer which results in a reactive capacity
requirement as shown in
Table 2.5, which is a corresponding requirement referred to the generator terminals [13].
Table 2.4: General requirement to reactive capacity for synchronous power-generating modules, referred to PCC
[13].

Requirements for type C and D, referred to 𝑃𝑚𝑎𝑥

Capacitive capacity 𝑄𝑐𝑎𝑝,𝑚𝑎𝑥 = 0.46 ∙ 𝑃𝑚𝑎𝑥 𝑐𝑜𝑠𝜑𝑐𝑎𝑝 = 0.91
Inductive capacity 𝑄𝑖𝑛𝑑,𝑚𝑎𝑥 = −0.46 ∙ 𝑃𝑚𝑎𝑥 𝑐𝑜𝑠𝜑𝑖𝑛𝑑 = 0.91
Requirements for type B, referred to 𝑃𝑚𝑎𝑥
Capacitive capacity 𝑄𝑐𝑎𝑝,𝑚𝑎𝑥 = 0.33 ∙ 𝑃𝑚𝑎𝑥 𝑐𝑜𝑠𝜑𝑐𝑎𝑝 = 0.95
Inductive capacity 𝑄𝑖𝑛𝑑,𝑚𝑎𝑥 = −0.33 ∙ 𝑃𝑚𝑎𝑥 𝑐𝑜𝑠𝜑𝑖𝑛𝑑 = 0.95

Table 2.5: General requirement to reactive capacity for synchronous power-generating modules, referred to
generator terminals, assuming 12 % reactive consumption in transformer [13].

Requirements for type C and D, referred to 𝑃𝑚𝑎𝑥

Capacitive capacity 𝑄𝑐𝑎𝑝,𝑚𝑎𝑥,𝑔 = 0.51 ∙ 𝑃𝑚𝑎𝑥 𝑐𝑜𝑠𝜑𝑐𝑎𝑝,𝑔 = 0.86
Inductive capacity 𝑄𝑖𝑛𝑑,𝑚𝑎𝑥,𝑔 = −0.33 ∙ 𝑃𝑚𝑎𝑥 𝑐𝑜𝑠𝜑𝑖𝑛𝑑,𝑔 = 0.95

2.4.3 Frequency requirements and energy reserves

Synchronous power-generating modules shall generally operate within the operating limits
described in Table 2.6, which states that the power-generating module shall not be limited
within the required time period. The requirement is vital to ensure a balanced power grid even
under large contingencies, as a too high imbalance may result in a total blackout.

Table 2.6: Requirements for the minimum operative period for synchronous power-generating modules, related
to frequency limits under the voltage range 0.9 – 1.05 Pu [13].
Frequency range Time period for operation
47.5 – 49.0 Hz 30 minutes
49.0 – 51.0 Hz Unlimited
51.0 – 51.5 Hz 30 minutes

One will achieve a balanced power grid or a constant frequency (50 Hz) when power
production and consumption are equal. In contrast, a grid exposed to a higher consumption will
have a lower frequency and vice versa when production is higher. According to the Nordic
25
 2 Operation of the Nordic grid
system operation agreement, the frequency in the Nordic grid should be kept within 50 ± 0.1
Hz [17]. Designating power-generating modules into different energy reserves allows the grid
to handle small and large grid imbalances with reduced risk. The reserves are also a part of the
Nordic balancing services, as discussed earlier. The reserves categorized into:
- Fast Frequency Reserves (FFR)
- Primary reserve: Frequency Containment Reserve (FCR)
- Secondary reserve: automatic Frequency Restoration Reserve (aFRR)
- Tertiary reserve: manual Frequency Restoration Reserve (mFRR)
Fast Frequency Reserves (FFR): Are reserves with a rapid response to large frequency
deviations, usually lower than 49.7 or 49.5 Hz [18]. The reserves shall secure the grid's stability
during significant faults and therefore activated within a second. The FFR is mainly a substitute
for the rotational inertia in power grids, where this is low.
Frequency Containment Reserve (FCR): The immediate change in load or production is
balanced by the rotating mass's energy in the power system before the frequency begins to
change. A frequency shift is handled first by the primary reserve FCR, an automatic controlled
reserve used to constrain the frequency change [19]. FCR distinguish between regular operation
(FCR-N), where the frequency is within a normal state and the process during disturbance
(FCR-D). The response time is usually within few seconds.
Automatic Frequency Restoration Reserve (aFRR): Also known as load frequency control
(LFC), where the reserves stabilise the frequency back to standard 50Hz. The TSO takes
automatically control over the generators in aFRR market with a response time within 2
minutes.
Manual Frequency Restoration Reserve (mFRR): When there is a significant imbalance in
the power grid or transmission bottlenecks, additional reserves will be activated manually by
the national TSO. This market is known as the capacity market (RK). Every participant offers
their mFRR price similar to the Elspot. During the activation, the cheapest reserves are first
activated, with a response time under 15 minutes. The response time and the relationship
between each energy reserve can be seen in Figure 2.9.

Figure 2.9: The response time of the different reserves [19].

26
 3 System overview

3 System overview
This theory chapter provides an overview of the construction and stationary losses associated
with a conventional hydropower unit (HPU). The main components which are covered are the
waterway, turbine, generator and transformer. This thesis focuses primarily on HPUs with
single generating units, i.e., HPUs with multiple generators, turbines, and waterways are not
considered.

3.1 Waterway
A waterway is a generic term for a river, tunnel, pipe, or other construction elements to convey
water from an upper to a lower reservoir. Waterway losses, often referred to as hydraulic losses,
are usually measured in the head2 [m] and proportional to the velocity squared, resulting in
substantial losses under high flow rates. A typical waterway efficiency is usually well above
90 % [20], referred to as nominal flow rate, but depends on the flow rate and the construction
of the waterway.

3.1.1 Construction
There are many components in a waterway construction, but the main components regarding
losses are the headrace/conduit, penstock, tail race, valves, and trash racks.
1. Headrace/conduits are usually a tunnel or a set of tunnels used to convey a large
amount of water, often over long distances of tens of kilometres, before connecting to
the penstock. It is possible to reduce water losses by having large horizontal tunnels as
the flow velocity become limited. In larger HPUs, tunnels will usually be blasting into
the mountains resulting in a rough surface if not smoothed out by, e.g., concrete. One
will often find one or more sand traps within the tunnels, as depicted in Figure 3.1. In
the sand traps, the flow velocity is relatively low, allowing the sand and rock sediments
to sink and accumulate at the bottom. The sand traps are essential, as sand and rock
sediments could damage the turbine.
2. Penstock is the pipe or tunnel connected to the turbine and is often a close to vertical
steel or concrete pipe with smooth surfaces, as depicted in Figure 3.2. The function of
the penstock is to convey water under high pressure and at a high velocity with limited
loss.
3. Tailrace is the tunnel or pipe which convey the water outflow from the turbine into the
lower reservoir or river. A tailrace will only add losses in fully submerged turbines, like
Francis and Kaplan, whereas a Pelton turbine would be unaffected.
4. Valves and trash racks are essential components in any conventional HPUs. When
an HPU is out of service, the valves may block water flow, whereas trash racks remove
any debris that may damage the turbine. Both valves and trash racks are small obstacles
in the waterway, which provides small additional losses.

2
Head is a synonym for height difference.

27
 3 System overview

Figure 3.1: Cleaning process of a sand trap in Rosekrepp hydropower [21].

Figure 3.2: A multiple penstock arrangement [22].

28
 3 System overview
3.1.2 Energy losses
As water moves through the waterway, energy loss, also known as hydraulic loss, accumulates
under geometric changes3 and friction forces in the water. The loss transforms the mechanical
energy of water into heat, which is unusable in hydropower. With turbulent and high flow
velocities, the magnitude of the loss will arise quickly. Hydraulic loss can be classified as [23]
[24]:

A typical illustration of head loss is seen in Figure 3.3, where the gross head (𝐻𝑔 ) [𝑚] is the
absolute difference in head between upper and lower reservoir, net head (𝐻𝑛 ) [𝑚] is the
available head for energy conversion and head loss (ℎ𝑓 ) [𝑚] is the energy loss.

Figure 3.3: Layout of an HPU, illustrating the relationship between the gross head, net head and head loss [24].

3
A change to the flow path like a bend or a change in cross section.

29
 3 System overview

The friction losses arise from the water viscosity, molecular and turbulent effects [23]. Friction
losses result from water molecules exchange their momentum, which occurs when molecules
in motion have different relative velocities. When the surface roughness is high, the water
becomes highly turbulent, causing friction losses to increase. A commonly used formula for
friction losses is Darcy-Weisbach’s equation [25] [26] [24]:
𝐿 ∙ 𝑣2
ℎ𝑓 = 𝑓 ∙ 3.1
2𝑔 ∙ 𝐷ℎ
where ℎ𝑓 = head loss [𝑚], 𝑓 = friction factor [𝑚], 𝑣 = flow velocity [𝑚/𝑠], 𝐿 = tunnel length
[𝑚], 𝑔 = gravitational constant [𝑚/𝑠 2 ] and 𝐷ℎ = hydraulic diameter [𝑚]. A friction factor is a
number describing the roughness of a pipe or tunnel, which can in some cases be obtained from
tables. In tunnels where the surface roughness is high, the friction factor could be measured,
like in the method proposed by Rønn and Skog [25], named IBA method. One can find more
details about friction losses in Appendix C.
Local losses arise in the flow path due to varying geometry or obstacles from swirling water;
this creates changes in the flow direction and localized pressure changes [23]. Local losses are
losses calculated from a single part or location and are therefore independent of the length.
Additional turbulence occurs whenever there is a change in the flow path, like pipe entrances
or exits, pipe bends, pipe contractions or expansions, see Figure 3.4 [26]. The equation for head
loss produced by local losses is given by [26]:
𝑘𝑣 2
ℎ𝑙 = 3.2
2𝑔
where ℎ𝑙 is the head loss in [m], 𝑣 is the flow velocity [𝑚/𝑠], g is the gravitational acceleration
[𝑚/𝑠 2 ] and k is the resistance coefficient for the pipe part/obstacle. The k-factor for different
pipe and valves can be found in tables as depicted in Appendix C.

Figure 3.4: Illustration of local losses produced in a pipe with sudden contraction and the rise of additional
turbulence [27]. “Vena contracta” is referred to as the point where fluid velocity is at its maximum.

30
 3 System overview
In practice, one will often come across waterway losses represented by a single head loss
coefficient (𝐾). The coefficient (K) is a number describing the total friction and local losses in
a waterway, given by the volumetric flow rate (𝑄𝑓𝑙𝑜𝑤 ). Volumetric flow rate, also called water
discharge, is regarded as the preferred unit as water velocity is relative to the pipe/tunnel size.
In contrast, the volumetric flow rate is constant throughout the waterway, assuming water is
incompressible. The equation for total head loss in a waterway based on the head loss
coefficient (K) yields:
2
𝐻𝑙𝑜𝑠𝑠 = 𝐾𝑄𝑓𝑙𝑜𝑤 3.3

Appendix D shows an example of the derivation of formula 3.3, based on equation 3.1 and 3.2.

3.2 Hydropower turbine

Hydropower turbines convert hydraulic power from the waterway into mechanical power,
which drives the generator. In conventional HPUs, there are three main types of hydro turbines,
called Pelton, Francis and Kaplan, where Francis is usually the turbine with the highest
efficiency, up to around 95 %, often referred to as the turbines best efficiency point (BEP). If
operated aside from the best point, it may reduce the efficiency significantly. Of course, the
operation and efficiency characteristics will depend on the type of turbine used, where typical
turbine characteristics are summarised [28] in Table 3.1.
Table 3.1: Typical specifications of turbines

Specification Pelton Francis Kaplan

Net head [𝑚] 200-2000 40-700 3-60
𝐹𝑙𝑜𝑤 𝑟𝑎𝑡𝑒 [𝑚3 /𝑠] 1-30 2-1000 30-1000
Operation [%] 5-100 50-100 20-100
Efficiency [%] <93 <96 <94

3.2.1 Main components – Francis and Kaplan turbines

A Francis and Kaplan turbine have almost identical construction, except the actual runner, as
depicted in Figure 3.5. The main components for Francis and Kaplan turbines are:
The spiral casing is the inlet to the turbine, where all the water from the reservoir is going
through. The design of the spiral casing is to guide the inlet water around the runner while
maintaining a constant velocity/pressure, achieved by an ever-decreasing cross-sectional area.
Guide vanes are fully adjustable blades used to maneuverer the water into the runner at an
optimal angle under an extensive range of operations with the additional ability to regulate the
flow rate into the turbine. The flow rate is the main parameter that determines the amount of
power the turbine will produce.

31
 3 System overview

Figure 3.5: The cross-section and the components of a) Francis turbine and b) Kaplan turbine
[29] [30].

The runner is the heart of the turbine, where the hydraulic power become converted to a
mechanical rotation that drives the generator. In the Francis turbine, the runner blades are fixed
and harness hydraulic power through lift and impulse forces. In comparison, the Kaplan turbine
has adjustable blades in addition to guide vanes, allowing Kaplan to have a larger operational
range than the Francis turbine. Kaplan turbines harness power only through lift forces.
The draft tube is only used by fully submerged turbines like Francis and Kaplan, connecting
the runner exit to the tailrace tunnel or the lower reservoir. After the water exits the turbine,
the backpressure will generally be less than atmospheric. By introducing an expanding tunnel,
called a draft tube, the high kinetic energy of the water become converted to a higher water
pressure that is useful as it increases the efficiency of the turbine before being discharged into
the tailrace.

3.2.2 Main components – Pelton turbines

The construction of a Pelton turbine has some differences from the structure of the Francis and
Kaplan turbine. The construction of a Pelton can be seen in Figure 3.7, and the main
components of the turbine are:
The distribution pipe is the pipe connecting the water inlet from the penstock to the nozzles.
It shares similar properties as the spiral casing, where the water pressure is maintained constant
at all nozzles.
The nozzles convert the potential energy of the water into pure kinetic energy, i.e., increasing
the velocity of the water as much as possible, resulting in a concentrated water jet. To regulate
the production, one could either adjust the nozzles' opening or choose the number of nozzles.
This feature allows excellent control of the flow rate while maintaining a constant water jet
velocity.

32
 3 System overview
Runners used in Pelton turbines have large set buckets, which converts the kinetic energy of
the water into mechanical energy for what is known as impulse forces. The buckets could
harness the hydraulic power by deflecting the water with an angle close to 180 degrees, as
illustrated in Figure 3.6. By having a peripheral runner speed close to half the jet velocity, the
exiting water jet will result in a limited velocity close to zero, relative to the surroundings,
resulting in high efficiency.

Figure 3.6: Cross-section of a Pelton bucket where the water jet reflects water at an angle of 165°. The figure is
based on a figure given in [31].

Figure 3.7: Construction of a five nozzle Pelton turbine [32].

33
 3 System overview
3.2.3 Energy losses – turbines
In a turbine, most energy losses are hydro-mechanical losses, categorised as [33]:
Friction losses occur in the distribution pipe, nozzles and buckets. There would be high friction
losses in the buckets due to the shear stresses between the buckets and the moving water
produced by what is known as viscous adhesion [33].
Windage losses are due to the air around the runner, which produces a drag on a rotating runner
Swirling losses are related to the velocity of the exiting flow out of the runner. The loss arises
from the existence of the rest kinetic energy that is present in the exit flow [33]. If the velocity
of the exit flow is zero, the swirling losses will also be zero. Swirling water could be server in
the draft tube of Francis and Kaplan turbines. Draft tube losses originate from swirling water,
also known as helical vortices, which produce a large pressure drop over the draft tube. The
minimal rotational motion of water exciting under BEP limits the size of the helical vortices,
as depicted in Figure 3.8. The figure shows the helical vortices in the part-load and full-loaded
condition in a test turbine [34].
Bearing losses are the losses produced by friction forces in the load-bearing to the runner.
Hydrodynamic plain bearings are almost exclusively applied to all Pelton turbines, as these can
handle high loads and stresses with minimal losses [33].
Erosion is an essential factor that may increase the friction losses after some years in service.
Erosion is widespread in locations where a large amount of sand is present or in turbines where
cavitation is present. Cavitation is a phenomenon produced by imploding water and occurs
when water is subjected to localized under pressure zones, typically found on Francis and
Kaplan turbines.
Leakage losses in Pelton turbines could be in the form of water that is not in contact with the
bucket, illustrated in Figure 3.9. In Francis and Kaplan turbines, some water escapes between
the runner and the sidewalls, thus wasting energy.
Impact losses are losses produced by water flowing in a direction that opposes the rotation of
the runner [28]. For a Pelton turbine, the deflected water may impact the runner and resist its
movement, whereas impact losses in Francis and Kaplan turbines occur at the runner's inlet.
One should have in mind that determining and make statements about the individual hydro-
mechanical are difficult as there has been hardly any available and reliable equations to make
it feasible with analytical and empirical formulas [33]. A common method is to use numerical
simulation software like computational fluid dynamics (CFD), which can differentiate
individual losses in turbines with high accuracies.

34
 3 System overview

Figure 3.8: Helical vortices formation in a draft tube at (a) part-load and (b) full-load regime
[34].

Figure 3.9: Energy loss in the form of leakage [35].

35
 3 System overview
3.3 Generator
A generator converts the mechanical power from the turbine into electrical power. The
efficiency of a large synchronous generator is in the order of 97-98.8% at nominal rating [20]
[36]. Synchronous machines are the preferred choice for generators larger than 1 MVA [37].
The reason is that the machines can have a four-quadrant operation and thereby regulate the
production of reactive power independently of the active power, making them superior to
asynchronous generator regarding stability capacity and voltage control. A typical hydropower
generator (salient pole) design is depicted in Figure 3.10.

Figure 3.10: Illustration of a salient-pole design of a large synchronous generator, used in an HPU [38]

3.3.1 Construction of synchronous machines

Synchronous machines come in two main types: cylindrical rotor and salient pole rotor,
illustrated in Figure 3.11. A cylindrical rotor type often found in high-speed applications, e.g.,
steam turbines or few cases, with Francis and Pelton turbines. However, these generator types
will not be discussed further. In contrast, salient pole rotor type machines are used for low-
speed (125-500 rpm) applications and almost exclusively used in all large HPU.
The synchronous generator consists of armature (stator) windings and field (rotor) windings
encased around an iron core made up from limitations of magnetic steel. Armature windings
conduct the primary current, i.e., the current transported out of the power plant and into the
grid. In comparison, the field windings conduct direct current to induce a magnetic field in the
rotor, allowing the machine to regulate the magnetic field strength or, more importantly,
reactive power production in the armature windings. With reactive power control, the generator
will be able to maintain the voltage.
An automatic voltage regulator (AVR) are used to control and excite the field windings. Since
the rotor is constantly rotating, the field windings are generally excited through what is known
as slip rings, where the electrical contact between two points are maintained while one of the
36
 3 System overview
points rotate. The slip rings usually have carbon (graphite) or metal brushes that slide or rubs
the surface of a metal ring located on the rotating part (rotor). However, the brushes produce
both friction and ohmic losses. Other technologies like brushless excitation systems use a
secondary generator to maintain electrical contact without slip rings, but this will not be further
discussed in this report.

Figure 3.11: Design of (a) cylindrical rotor and (b) salient pole rotor [39].

3.3.2 Machine losses

In a generator, there are load-dependent losses considered variable, and there are constant
losses that are independent of the loading. The IEEE standard [15] have a general categorisation
of machine losses which is as follows:

Winding losses
The ohmic resistance in windings (stator and rotor) produces what is known as winding losses,
also referred to as 𝐼 2 𝑅 loss. The total winding loss in the stator is the sum of the winding
resistance for each phase, expresses as:
𝑃𝑆 = 3 ∙ 𝐼 2 ∙ 𝑅𝑠,𝑠𝑡𝑎𝑡𝑜𝑟 3.4

37
 3 System overview
On the other hand, the field winding losses have additional losses considering the total
excitation system (e.g., AVR). One can assume the excitation losses to be a linear function of
the field current (𝐼𝑓𝑑 ), with a contribution of about 10% (k = 1.1) that are added to the field
winding losses [40] and [41], yielding:
2
𝑃𝑅 = 𝑘 ∙ 𝐼𝑓𝑑 ∙ 𝑅𝑠,𝑟𝑜𝑡𝑜𝑟 3.5

where 𝑅𝑠,𝑟𝑜𝑡𝑜𝑟 [Ω] is the measured resistance (including bush resistance), 𝑃𝑅 is the total rotor
winding loss, 𝑘 is the proportional constant for excitation system loss. 𝐼𝑓𝑑 is the field current
and is usually given in the form of open-circuit characteristic (OCC) and air-gap characteristic.
The air-gap characteristic represents the relationship between excitation voltage and field
current when the core unsaturated. The OCC describes the relationship between excitation
voltage and field current in both unsaturated and saturated situation. A calculation example of
field current is found in Appendix E.
According to IEEE STD 115 [42], the winding resistance (𝑅𝑠,𝑠𝑡𝑎𝑡𝑜𝑟 and 𝑅𝑠,𝑟𝑜𝑡𝑜𝑟 ) should be
given at specified temperature (normally 75 ℃). The formula for winding resistance under a
given temperature, is expressed as:
𝑅𝑠 = 𝑅0 (1 + 𝛼𝑡 (𝑇𝑠 − 𝑇0 )) 3.6
where
𝑅𝑠 is the winding resistance [Ω], corrected to the specified temperature, 𝑇𝑠
𝑇𝑠 is the specified temperature [℃]
𝑅0 is the nominal winding resistance [Ω]
𝑇0 is the nominal temperature [℃] of winding when resistance was measured, usually
20[℃]
𝛼𝑡 is the temperature coefficient for the winding material (0.00386 for pure copper or
0.00429 for aluminium)

Core losses
A generator with an applied voltage that is alternating (AC) will produce core losses. The
induced alternating magnetic fluxes in the iron core produces what is known as hysteresis and
eddy currents. These effects will then generate core losses in the form of heat. One can often
assume the core losses to be constant by having minor voltage variations. In reality, the core
loss is voltage-dependent, as depicted in Figure 3.12. It is possible to estimate the core loss by
measuring the change in power draw with and without excitation under an open circuit operated
at the constant nominal speed [15].

38
 3 System overview

Figure 3.12: Armature core loss relative to the applied terminal voltage, as described in the IEEE STD 115 [42].

Over the years, numerous calculation methods have been proposed, like the well-known
Steinmetz’s equation, published by Charles Steinmetz in 1892 [43] or newer methods as shown
in the study from Ionel [44]. Most core loss expressions estimate the energy loss based on the
friction, flux density and material coefficients. However, advanced field simulations software
like FEM (finite element method) are often required to achieve accurate results.
Hysteresis is a phenomenon observed in ferromagnetic materials like steel subjected to an
alternating magnetic field. In ferromagnetic materials, the molecular structure can rotate to an
adjacent field, and this motion requires energy, resulting in energy loss or heat. The polarisation
of the structure is often illustrated in a hysteresis loop, as seen in Figure 3.13. The hysteresis
diagram describes the relationship between the induced magnetic flux density (B) and the
magnetisation force (H), often referred to as B-H loop [45].

Figure 3.13: Illustration of a hysteresis loop [46].

Eddy currents are due to the induced circular currents conducted in the laminated magnetic
steel that produces ohmic losses. Creating a core with laminated magnetic steel will decrease
the electrically conductive path making a higher electrical resistance and reducing the induced
eddy currents.

39
 3 System overview
Stray-load loss
Under loaded conditions, the armature current induces an alternating magnetic field with a field
strength directly proportional to the current, resulting in additional eddy currents in the
armature windings and the iron core, often referred to as stray-load loss.
Copper stray-load loss: Alternating magnetic fields produces eddy currents inside the
windings and give rise to a non-uniform current distribution in the windings. An effect is
known as the skin-effect and proximity effect. The non-uniform current distribution decreases
the effective cross-sectional area, resulting in increased resistance and ohmic losses
proportional to the loading current squared, see Figure 3.14.

Figure 3.14: Armature winding loss and stray-load loss, as described in the IEEE STD 115
[42]
Core stray-load loss: Under the loaded condition, the iron core and teeth are highly saturated,
allowing more flux to leak through the core that produces losses in the form of eddy currents
in the shielding, stator cover and end frames [47].
Modelling stray-load losses are complex, and FEM simulation tools are often required, but the
losses could be measured as described in the IEEE Std C50.13 [15].

Mechanical losses
Friction loss is the friction produced between the rotor shaft and generator bearings and the
friction between the slip rings and brushes.
Windage loss (or viscous friction) is caused by air friction when the rotor rotates and is
proportional to the rotor velocity cubed. A first approximation formula for windage losses has
been proposed by J. Vrancik [48] in 1968, where the equation calculates the windage loss for
a smooth cylinder rotating within a concentric cylinder and corrects for a salient pole design.
For more detailed estimations, a CFD simulation would be required.
Ventilation and cooling losses are the power required to cool down the generator system,
containing fans and circulation pumps [15].

40
 3 System overview
3.3.3 Capability diagram
The operational boundaries to a generator are defined by the P-Q plane's limits, also known as
the capability diagram. The capability diagram describes the maximum active and reactive
power allowed for the generator to operate concerning thermal limits and stability limits under
a constant (nominal) voltage. The limiting factors in the capability diagram are illustrated in
Figure 3.15 and are given by:
- Armature current limit (c-d) and (e-f) produced by ohmic 𝐼 2 𝑅 losses in the armature
windings, where the upper temperature in the windings are the limiting factor.
- Upper prime mover limit (d-e) is determined by the maximum active power the HPU
could continuously deliver, usually rated by the turbine.
- The lower prime mover limit (a-g) is the minimum power the turbine could produce
continuously. A lower limit is only applicable for Francis and Kaplan turbines which
have a limit of around 5-30 % of the rated output [49].
- Field current limit (f-g) produced by ohmic 𝐼 2 𝑅 loss in the field windings, where the
upper temperature in the windings are the limiting factor.
- End heating limit (a-b), localised heating in the end region of the stator laminations
produced by electromagnetic fields entering and exiting perpendicular to the
laminations [50]. In under excited operation, the field current is low, and the so-called
retaining ring will not be saturated, resulting in high flux leakages. As a consequence,
more eddy currents will be induced in the laminations that will be added to the eddy
currents produced by the armature currents, and there will be excessive heating at the
end laminations.
- Practical stability limit (b-c), a limit used to reduce the risk of losing synchronism, also
known as skipping, when the generator is situated in an under-magnetised operation.

41
 3 System overview

Figure 3.15: Illustration of a capability diagram to a salient pole generator.

3.4 Transformer
A transformer is an electrical component used to increase the relatively low voltage induced
by the generator to reduce transmission losses. The transformer is a relatively simple
component; it contains a primary and secondary coil wound around an iron core, see Figure
3.16. The ratio between the number of windings determines the ratio to which the voltage will
increase/decrease. The transformer is the component in an HPU with the highest efficiency,
usually close to 99.7% at nominal power. Due to the extraordinary high efficiency of the
transformer, a simplified calculation model is seen as adequate and will only be described
briefly.

3.4.1 Transformer losses

A transformer has a lot in common with generators when it comes to theoretical background
and losses. In simplified terms, the transformer losses can be represented by no-load losses (𝑃0 )
and load losses (𝑃𝑘 ) [46].
No-load losses (𝑃0 ), often referred to as magnetisation losses or core losses will be present as
long as the transformer is energized. The no-load losses are usually assumed to be constant by
not considering any voltage differences and flux leakages [46]. The magnetisation losses are
produced by hysteresis and eddy currents, as described in section (0).
42
 3 System overview
Load losses are produced under loaded conditions, where the load current and the resistance
result in ohmic losses (𝐼 2 𝑅). The load-dependent losses will, in addition, be affected by
temperatures and stray-load losses, as described in section (0). For simplicity, one can represent
the load losses as a proportional constant (𝑃𝑘 ) that are multiplied to the loading ratio, seen in
equation 3.7.
By combining no-load losses (𝑃0 ), and load losses (𝑃𝑘 ), the total transformer loss (𝑃𝑙𝑜𝑠𝑠 ) and
efficiency (𝜂𝑇𝑟𝑎𝑛 ) can be expressed as:

𝑆𝑙𝑜𝑎𝑑 2
𝑃𝑙𝑜𝑠𝑠 = 𝑃0 + 𝑃𝑘 ( ) 3.7
𝑆𝑛
𝑃0
𝜂𝑇𝑟𝑎𝑛 = 3.8
𝑃0 + 𝑃𝑙𝑜𝑠𝑠
where 𝑆𝑙𝑜𝑎𝑑 is the load out of the transformer [MVA or Pu] and 𝑆𝑛 is the transformer’s rating
[MVA or Pu].

Figure 3.16: Design of a large power transformer (ABB’s TrafoStar design) [51].

43
 4 Power plant description

4 Power plant description

In this chapter, there will be a description of the two hydropower units analysed, namely Åbjøra
and Sundsbarm. Both of the units are regulated impoundment types with a single generating
unit. Åbjøra and Sundsbarm are owned or partly owned by Skagerak Kraft and are located in
Nord-Aurdal and Seljord. A summary of the nominal specifications to Åbjøra and Sundsbarm
can be seen in Table 4.1. The specifications depicted in Table 4.1 form the basis for most of
the calculations used in the analysis. Additional details considering the generator and turbine
characteristics relevant for simulation is found in Appendix F (Åbjøra) and Appendix G
(Sundsbarm).

Table 4.1: Overview of the nominal specifications to Åbjøra and Sundsbarm hydropower station. * Mechanical
losses in Sundsbarm contains windage, ventilation and bearing losses.

Åbjøra Sundsbarm Specification

Value Value
442 m 493 m Nominal gross head
24 𝑚3 /𝑠 24 𝑚3 /𝑠 Nominal discharge
5140 m 7750 m Waterway length
550 GWh 430 GWh Average annual production
0.016335 0.00893 Total head loss coefficient K (waterway)
95 MW 103MW Nominal turbine power
94.26 % 93.86 % Nominal turbine efficiency
103 MVA 118 MVA Nominal generator power (𝑐𝑜𝑠𝜑 = 0.9, 0.85)
98.8 % 98.6 % Nominal generator efficiency
0.15253 Ω 0.24857 Ω Field winding resistance (75℃)
0.003155 Ω 0.00390 Ω Armature winding resistance (75℃)
11 kV 15 kV Nominal voltage
172.92 kW 537 kW* Mechanical losses (windage and ventilation)
211.92 kW 353 kW Core losses
240.90 kW − Bearing losses (generator)
276.62 kW 333 kW Nominal load losses
191.66 kW 226 kW Nominal total excitation losses
1094.02 kW 1450 kW Nominal total losses
103 MVA 118 MVA Nominal transformer rating
52.5 kW 148.4 kW No-load losses (𝑃0 )
225 kW 281.4 kW Load losses (𝑃𝑘 )

44
 4 Power plant description
4.1.1 Åbjøra
Åbjøra hydropower station is located in Nord-Aurdal, Norway and owned by Skagerak Kraft.
The station was first opened in the year 1951 with three Pelton turbines with a total rating of
81 MW. Åbjøra was later rebuilt in 2002, where all three Pelton turbines were replaced (Figure
4.1) by a single Francis turbine (Figure 4.2) with a rating of about 95 MW with an average
annual production of 550 GWh. The increased power resulted from a more efficient turbine
and changes to the waterway, as the new station was placed 15 meters below the old station.
The hydropower utilizes multiple reservoirs joined to an intake reservoir at Bløytjern/Ølsjøen,
which results in a nominal gross head of 442 m. Thought a year, the gross head only varies
from about 439 m to 445 m or less.

Figure 4.1: Disassembly of the old Pelton turbines in Åbjøra (2007). [Photo: Jan Erik Olsrud]

45
 4 Power plant description

Figure 4.2: Inside of Åbjøra power station, showing the generator. [Photo: Skagerak Kraft]

In the waterway, there is a 4.5 km headrace tunnel from the intake reservoir before entering a
vertical pressure shaft of 380 m followed by an additional pressure shaft of 260 m before
entering the turbine. After the rebuild in 2002, the losses in the waterway were measured. The
measurements could be seen in Table 4.2, which shows the head loss coefficients (𝑘𝑖 ) for each
respectful element in the waterway, see Figure 4.3. The total head loss coefficient (𝐾) of the
waterway was estimated to 𝐾 = 0.016335.

Figure 4.3: Simplified overview of the waterway in Åbjøra, where the numbers (1-9) represents the different
elements used in the measurements found in Table 4.2. In the figure one can see the conduit and penstock is
divided into two whereas the tailrace is divided into three elements.

46
 4 Power plant description

Table 4.2 Head loss measurements of each element in the waterway from Åbjøra
Waterway elements and their respectful head loss coefficients
Object Element Element nr. Head loss coefficient (𝑘𝑖 ) Power loss [%]
Conduit (tunnel part 1) 1 0.01
61.8
Conduit (tunnel part 2) 2 0.00009
Penstock (part 1) 3 0.00069
5.7
Penstock (part 2) 4 0.000245
Main valve + turbine 5 0.0022 13.5
Draft tube 6 0.00011 0.7
Tailrace (tunnel part 1) 7 0.0001
Tailrace (tunnel part 2) 8 0.00015 18.4
Tailrace (tunnel part 3) 9 0.00275
Total 𝟎. 𝟎𝟏𝟔𝟑𝟑𝟓 100

The Francis turbine operates under a nominal gross head of 442 m (net head of 433m) and a
nominal discharge of about 24 𝑚3 /𝑠. Measurements from 2003 gave the results shown in
Appendix F, which indicate a nominal efficiency of 94.26 % and a maximum efficiency of
94.56 % achieved around 86 MW under the nominal head. It has been assumed that the turbine
efficiency correlates to the runner efficiency as the data includes both the main valve and
turbine (spiral casing and guide vane) and draft tube as a part of the waterway.

4.1.2 Sundsbarm
Sundsbarm hydropower station is located in Seljord, Norway and is owned by Skagerak Kraft
AS (91.5%). The station was put into operation in 1970 with a Francis turbine of 103 MW,
depicted in Figure 4.4. Sundsbarm is a reservoir type hydropower, which uses the lake named
Sundsbarmvatnet as a reservoir and utilizes few additional lakes interconnected through
tunnels, resulting in a precipitation area of about 418 𝑘𝑚2 . Throughout a year, the gross
headwater could range from 455 m to 493 m, with and an annual production of about 430 GWh.
From the intake reservoir in Sundsbarmvatnet, the water in guided through a 6.6 km conduit
tunnel before entering a vertical pressure shaft of 600 m. A simplified overview of the
waterway is shown in Figure 4.5. In 2011, the head loss in the waterway was measured, these
results can be seen in Table 4.3, which show the head loss coefficients for each respectful
element in the waterway (Figure 4.5). The total head loss coefficient for the waterway is
estimated to be 𝐾 = 0.00893.
Table 4.3: Head loss measurements of each element in the waterway from Sundsbarm.
Waterway elements and their respectful head loss coefficients
Object Element Element nr. Length [m] Head loss coefficient (𝑘𝑖 ) Power loss [%]
Conduit (tunnel part 1) 1 3000 0.0026
61.6
Conduit (tunnel part 2) 2 3600 0.0029
Penstock 3 600 0.00295 33.0
Draft tube and Tailrace 4 550 0.00048 5.4
Total 7750 0.00893 100

47
 4 Power plant description

Figure 4.4: Inside of Sundsbarm power station, showing the generator. [Photo: Skagerak Kraft]

Figure 4.5: Simplified overview of the waterway in Sundsbarm. The element numbers (1-4) represents the
object element found in Table 4.3.

48
 4 Power plant description
The Francis turbine is operated under a nominal gross head of 480 m (net head of 487 m) and
a nominal discharge of about 24 𝑚3 /𝑠. Estimates from 2011 gave the results shown in
Appendix G, which indicate a nominal efficiency of 93.86 % and a maximum efficiency of
93.93 % achieved around 100 MW. Similar to Åbjøra, it has not been specifically defined what
the turbine measurements evaluate, but it is assumed it contains only the runner efficiency.
The generator is a synchronous generator with nominal apparent power of 118 MVA and PF =
0.85 and a nominal voltage of 15 kV. In Appendix G, one can find the generator’s load
characteristics and measurements.

49
 5 Methods

5 Methods
This chapter contains descriptions of the methods used to solve the analysis objectives. The
main tool of this analysis is the simulation model of the HPU, created in the software
MATLAB. The simulation model is used to map and assess possible optimal efficiency, also
known as the best efficiency point (BEP). In addition, the model will be used for analysing the
operational efficiencies and energy losses. The model uses input parameters from the
waterway, turbine, generator and transformer, and uses operational data taken from the
transformer (high voltage side) or generator to determine the losses in all of the components.
The basic structure of the simulation model is illustrated in Figure 5.1, which depicts all input
parameters and variables, main functions and output variables. The input parameters are shown
with their respective unit where data acquisition represents the filtered input variables, active
power (P), reactive power (Q) and voltage (V).

Figure 5.1: Diagram describing the structure of the simulation model.

50
 5 Methods
5.1 Simulation model
In this report, the simulation (HPU) model is a general mathematical model coded in MATLAB
and is used to calculate energy losses and efficiencies under static conditions given by
operational data of active power (P), reactive power (Q) and voltage (V). The model has built-
in features which allow an additional input of gross head (𝐻𝑔𝑟𝑜𝑠𝑠 ) to be added, but will not be
used in this report as there was no available operational data about the head. The HPU model
is a collection of several models (transformer, generator, turbine and waterway) which are
combined into a single HPU model, illustrated in Figure 5.2. The figure depicts how each
model is interconnected with its respective input parameters, variables and outputs.

Figure 5.2: Structure of the HPU model.

The HPU model uses the production data from the transformer (or generator), as this data is
widely accessible for most HPUs. Thus, allowing easy implementation of the model to, e.g.,
more extensive grid simulations. Knowing the efficiency of every component and having the
production output of the transformer or generator accessible, one can simply predict the power
flow through the turbine and the waterway and then determine the water or power consumption.
Calculating from the production side to the input side causes the models closest to the
production side (transformer) to transfer the least amount of power. However, the models
furthest away (waterway) from the production side will transfer the most power as the losses
for each model are added along the way, illustrated in Figure 5.3.

Figure 5.3: Power flow diagram, illustrating the calculation direction, where Pi represents the electrical
production (input data) of the transformer and Po represents the hydraulic power entering the waterway (output
data from the model).

51
 5 Methods
A single mathematical function can express how the efficiency of the HPU model and each
sub-model is calculated, which yields:
𝑃
𝜂𝐻𝑃𝑈 (𝑃𝑙𝑜𝑎𝑑 , 𝑄𝑙𝑜𝑎𝑑 , 𝑉𝑙𝑜𝑎𝑑 , 𝐻𝑔𝑟𝑜𝑠𝑠 ) = 𝜂𝑇𝑟𝑎𝑛 (𝑃𝑙𝑜𝑎𝑑 , 𝑄𝑙𝑜𝑎𝑑 ) ∙ 𝜂𝐺𝑒𝑛 (𝜂 𝑙𝑜𝑎𝑑 , 𝑄𝐺𝑒𝑛 , 𝑉𝐺𝑒𝑛 ) ∙
𝑇𝑟𝑎𝑛
𝑃𝑙𝑜𝑎𝑑 𝑃𝑙𝑜𝑎𝑑 5.1
𝜂𝑇𝑢𝑟 (𝜂 , 𝐻𝑔𝑟𝑜𝑠𝑠 ) ∙ 𝜂𝑊𝑎𝑡𝑒𝑟 (𝜂 , 𝐻𝑔𝑟𝑜𝑠𝑠 )
𝑇𝑟𝑎𝑛 ∙𝜂𝐺𝑒𝑛 𝑇𝑟𝑎𝑛 ∙𝜂𝐺𝑒𝑛 ∙𝜂𝑇𝑡𝑢𝑟

where 𝜂𝐻𝑃𝑈 is the total plant efficiency, 𝜂𝑇𝑟𝑎𝑛 is the transformer efficiency, 𝜂𝐺𝑒𝑛 is the
generator efficiency, 𝜂𝑇𝑢𝑟 is the turbine efficiency, and 𝜂𝑊𝑎𝑡𝑒𝑟 is the waterway efficiency.
𝑃𝑙𝑜𝑎𝑑 [𝑀𝑊], 𝑄𝑙𝑜𝑎𝑑 [𝑀𝑉𝐴𝑟], 𝑉𝑙𝑜𝑎𝑑 [𝑘𝑉] are the input (operational data) at the transformer’s high
voltage terminals and 𝐻𝑔𝑟𝑜𝑠𝑠 [𝑚] is the gross head given at nominal value.
For this report, the operational data was measured from the generator terminals and not the
high voltage terminals of the transformer, which the HPU model is designed for. Thus, the
operational data was first adjusted by subtracting the energy loss of the transformer before
implemented into the model, explained in detail in section 5.6. Since the inputs were given at
the generator terminals, the absorption of reactive power in the transformer could be neglected,
i.e., 𝑄𝑙𝑜𝑎𝑑 = 𝑄𝑔𝑒𝑛 . In addition, due to the lack of data, the gross head was assumed constant
and was set to nominal gross head for all operations. All parameters used for Åbjøra and
Sundsbarm model is summarised in Appendix F and Appendix G, respectfully. A function
description of the different functions used in the MATLAB model can be found in Appendix
K. In addition, the MATLAB code all functions, including the code for Åbjøra with various
plot functions, can be found in Appendix L.

5.2 Data acquisition and preparation

The operational data originates from the system called SCADA “Supervisory Control and Data
Acquisition” and is the system used for measuring and transport information to the control
centre [52], in this case, Skagerak Kraft. This data contributes to analysing the operational
efficiency and energy losses of the respective hydropower plants.

5.2.1 Data format

Skagerak Kraft provided the data in CSV (Comma-Separated Values) format, a typical file
format used to store measurements, where a comma separates each column. CSV files are
plain-text files, which makes them compatible with most analysis/data programs and allow for
easy transport of tables/measurements between different programs.
The data contained timestamps (date and time) and generator measurements of the active power
[MW], the reactive power [MVAr] and the voltage [kV], given as average values over 1 hour
periods. The operational data shows the operation period between (22.01.2020 – 31.12.2020),
i.e., a total number of 8256 hours or data points.

52
 5 Methods
5.2.2 Collecting and reading the CSV-files
CSV files are practical for transporting and storing measurements but are impractical for
visualization and working with the data. Thus, the files was first converted to an excel-format
with the help of the excel-functions named “From text/CSV” or “Text to columns”. Before the
data could be implemented into MATLAB for analysis, the data file was filtered and corrected
due to:
- Measurements with equal repetitive values were represented as “NaN” which needed
to be replaced with actual values.
- Data were given as average values, which results in measurements of active power (P),
reactive power (Q) and voltage (V) with highly inaccurate values that do not reflect
actual operating regimes.
- The combination of measurements with average values and the repetitive values given
as “NaN” resulted in a data set that did not differentiate between operation in the “on”
and “off “state.

5.2.3 Data processing

The procedure for formatting and correcting the data was performed in excel and described in
four steps where the original data set is gradually improved until the finished format. The full
description with excel commands can be seen in Appendix H. The summary of each step goes
as follow:
- Step 1: Convert all operating points marked “NaN” with actual numbers by assuming
each “NaN” is equal to the previously known value.
- Step 2: Corrected for logical errors where it seems like the generator have multiple
start/stop sequences between two stable regions, and these regions will be set to zero.
A stable region is regarded as a region where the voltage is above a trigger point or a
voltage that is considered close to “normal”.
- Step 3: Convert data from 1-hour resolution to 15-minute semi-resolution by copying
each operation four times for differentiating between the “on” and “off” state during
the start/stop region.
- Step 4: Estimate the data in the 15-minute semi-resolution. The duration of the on-state
will be determined by the ratio between the operating voltage and trigger (“normal”)
voltage. Assumptions during a start/stop sequence:
o A voltage or active power value above or equal a maximum trigger value will
be unchanged.
o A voltage or active power below a maximum trigger point and active power
above a minimum trigger value will be set to a chosen value that is considered
normal.
o A voltage or active power below a minimum trigger value will be regarded as
an”off” state and set to zero.

53
 5 Methods
o Reactive power will not be corrected as there is no “normal” operating point
and can be both positive and negative.

An illustration of the filtering process is seen in Table 5.1, showing the result of each step. The
illustrated example is taken from Sundsbarm, depicting the transition between full operation to
off-state. Within the given period, 116 and 55 off-transitions have been identified, which
accounts for about 1.4 % and 0.7 % of the total operations for Åbjøra and Sundsbarm,
respectfully. Due to the assumptions stated in step 4, the filtering process may result in few
incorrect evaluations. Still, with the low number of situations (1.4 % and 0.7 %), the effect
could be assumed insignificant, in particular, to the energy loss and average efficiency.

Table 5.1: Example of the filtering process of the voltage to Sundsbarm. The “normal” value or the maximum
trigger voltage is here chosen to be 14.4 kV. In operation nr. 2, the 𝑑𝑢𝑟𝑎𝑡𝑖𝑜𝑛 = 9.59/14.4 = 0.67 or rounded
to 3/4, which can be seen in step 4.
Original Step 1 Step 2 Step 3 Step 4 (Finished)
14.3667352 14.3667352
14.3667352 14.3667352
14.3667352 14.3667352 14.3667352
14.3667352 14.3667352
14.3667352 14.3667352
9.5889163 14.4 (trigger voltage)
9.5889163 14.4 (trigger voltage)
9.5889163 9.5889163 9.5889163
9.5889163 14.4 (trigger voltage)
9.5889163 0
0 0
0 0
NaN 9.5889163 0
0 0
0 0

5.3 Waterway model

A single equation will represent the waterway model. The equation estimates the waterway
efficiency and power loss based on the head loss coefficient (𝐾) and the volumetric flow
(𝑄𝑓𝑙𝑜𝑤 ). From theory (section 3.1), it was claimed that the waterway is generally represented
by the head loss [m]. However, describing the losses in terms of active power [MW] would be
more practical for this model regarding compatibility between the other models.
For this report, the only known parameters for the waterway was the gross head (𝐻𝑔𝑟𝑜𝑠𝑠 ), the
power output of the waterway (𝑃𝑤𝑎𝑡𝑒𝑟𝑤𝑎𝑦,𝑜𝑢𝑡 ) and the head loss coefficient (K). Under these
conditions, a method was developed for determining the power loss [MW] in the waterway.
The method is solely based on the standard hydropower formulas for hydraulic power plants.
The method is as follows:
The total available power (output) of the waterway is [28]:
𝑃𝑤𝑎𝑡𝑒𝑟𝑤𝑎𝑦,𝑜𝑢𝑡 = 𝜌𝑔𝑄𝑓𝑙𝑜𝑤 𝐻𝑛 5.2

54
 5 Methods
where 𝐻𝑛 = net head [m], 𝜌 = water density and g = gravitational constant. Both 𝑄𝑓𝑙𝑜𝑤 and 𝐻𝑛
are unknown variables. 𝑃𝑤𝑎𝑡𝑒𝑟𝑤𝑎𝑦,𝑜𝑢𝑡 will be known, as this power is given by the turbine.
From equation 3.6:
2
𝐻𝑙𝑜𝑠𝑠 = 𝐾𝑄𝑓𝑙𝑜𝑤 5.3

Substituting 𝐻𝑛 in (5.2) with gross head (𝐻𝑔𝑟𝑜𝑠𝑠 ) and head loss (𝐻𝑙𝑜𝑠𝑠 ) (5.3) gives:
2
𝑃𝑤𝑎𝑡𝑒𝑟𝑤𝑎𝑦,𝑜𝑢𝑡 = 𝜌𝑔𝑄𝑓𝑙𝑜𝑤 (𝐻𝑔𝑟𝑜𝑠𝑠 − 𝐾𝑄𝑓𝑙𝑜𝑤 ) 5.4

Rearrange (5.4) gives the expression of a cubic equation:

3
𝑃𝑤𝑎𝑡𝑒𝑟𝑤𝑎𝑦,𝑜𝑢𝑡
𝐾𝑄𝑓𝑙𝑜𝑤 − 𝐻𝑔𝑟𝑜𝑠𝑠 𝑄𝑓𝑙𝑜𝑤 + =0 5.5
𝜌𝑔
Solving the equation with respect to discharge 𝑄𝑓𝑙𝑜𝑤 , gives the flow rate required for yielding
the output power under a given gross head and head loss coefficient. With the estimated
discharge 𝑄𝑓𝑙𝑜𝑤 , the power loss 𝑃𝑙𝑜𝑠𝑠 produced in the waterway is then calculated by using
(5.2) and substituting 𝐻𝑛 with head loss (5.3):
3
𝑃𝑙𝑜𝑠𝑠 = 𝐾𝜌𝑔𝑄𝑓𝑙𝑜𝑤 5.6

Substituting (5.2) and (5.6) yields the total waterway efficiency:

𝑃𝑜𝑢𝑡 𝑃𝑤𝑎𝑡𝑒𝑟𝑤𝑎𝑦,𝑜𝑢𝑡 𝑃𝑤𝑎𝑡𝑒𝑟𝑤𝑎𝑦,𝑜𝑢𝑡
𝜂𝑊𝑎𝑡𝑒𝑟 = = = 3 5.7
𝑃𝑖𝑛 𝑃𝑤𝑎𝑡𝑒𝑟𝑤𝑎𝑦,𝑜𝑢𝑡 + 𝑃𝑙𝑜𝑠𝑠 𝑃𝑤𝑎𝑡𝑒𝑟𝑤𝑎𝑦,𝑜𝑢𝑡 + 𝐾𝜌𝑔𝑄𝑓𝑙𝑜𝑤

5.4 Turbine model

The turbine model will be represented by either a simple 1D interpolation or a 2D interpolation.
Interpolation is an ordinary mathematical operation used to predict the value or line which is
the most probable between two or more data sets. For the turbine, it has been provided
operational data of the output power (𝑃) and efficiency (𝜂𝑇𝑢𝑟 ). With two data sets (𝑃) and
(𝜂𝑇𝑢𝑟 ), one will have what is known as a 1D interpolation (linear, cubic or higher order),
illustrated in Figure 5.4, where the blue circles and the orange curve represents the known data
points and the interpolated curve, respectfully. MATLAB has implemented a simple function
for 1D interpolation named “cubic spline” or simply “spline”, used in this report.
In the turbine model, bilinear interpolation (2D) have been added as an extended feature to the
software. The bilinear interpolation allows the user to interpolate between output power (𝑃),
efficiency (𝜂𝑇𝑢𝑟 ) and gross head (𝐻𝑔𝑟𝑜𝑠𝑠 ), which results in a three-dimensional plot, as depicted
in Figure 5.5. Bilinear interpolation is an extension of linear interpolation where linear
interpolation is first performed between, e.g., (𝑃) and (𝜂𝑇𝑢𝑟 ) for all known heads (𝐻𝑔𝑟𝑜𝑠𝑠 ).
Then, bilinear interpolation will be achieved by interpolating again, but this time between all
the interpolated curves.
55
 5 Methods
The turbine model has the 1D interpolation implemented, which can be used for cases where
one have only measurements from one head. In this report, the turbine measurements were
given for multiple head, allowing 2D interpolation to be implemented.

Figure 5.4: An illustration of the efficiency curve of a Francis turbine, where the dotted line is the given data
points, and the red line is the interpolated curve.

Figure 5.5: Efficiency mapping of Sundsbarm, with bilinear interpolation between active power (𝑃), gross head
(𝐻𝑔𝑟𝑜𝑠𝑠 ) and total plant efficiency (𝜂𝐻𝑃𝑈 ).

56
 5 Methods
5.5 Generator model
The generator model is based on the work form [40], and the IEEE standard [15], where the
total energy loss in the generator is expresses as:
𝑃𝑙𝑜𝑠𝑠 = 𝑃𝑆 + 𝑃𝑅 + 𝑃𝐹𝐸 + 𝑃𝑉 + 𝑃𝐹 5.8
where 𝑃𝑆 is winding loss of stator windings, 𝑃𝑅 is winding loss of rotor windings and excitation
losses, 𝑃𝐹𝐸 is iron core loss, PV is the ventilation and cooling loss, and PF is friction and windage
loss. Thereby, the total generator efficiency becomes:
𝑃𝑙𝑜𝑎𝑑
𝜂𝑡𝑜𝑡 = 5.9
𝑃𝑙𝑜𝑎𝑑 + 𝑃𝑙𝑜𝑠𝑠 (𝑃𝑡 , 𝑄𝑡 , 𝑉𝑡 )
where 𝑉𝑡 is the terminal voltage of the generator, 𝑃𝑡 and 𝑄𝑡 is the active and reactive power
produced by the generator, respectfully.
In this report, the iron core loss (𝑃𝐹𝐸 ), ventilation and windage (𝑃𝑉 ) and friction loss (𝑃𝐹 ) has
been assumed constants for all operating regimes. These constant losses were fortunately
provided, meaning the stator (𝑃𝑆 ) and rotor (𝑃𝑅 ) losses were the only losses that needed to be
estimated. In the generator model, it was used the same equations for stator (𝑃𝑆 ) and rotor (𝑃𝑅 )
losses, as described in section (3.3), which yields:
𝑃𝑆 = 3 ∙ 𝐼 2 ∙ 𝑅𝑠,𝑠𝑡𝑎𝑡𝑜𝑟
2
𝑃𝑅 = 𝑘 ∙ 𝐼𝑓𝑑 ∙ 𝑅𝑠,𝑟𝑜𝑡𝑜𝑟
where 𝐼 is the armature/phase current, 𝐼𝑓𝑑 is the field current, 𝑅𝑠,𝑠𝑡𝑎𝑡𝑜𝑟 is the armature/phase
winding resistance and 𝑅𝑠,𝑟𝑜𝑡𝑜𝑟 is the field winding resistance. In this report, the 𝑅𝑠,𝑠𝑡𝑎𝑡𝑜𝑟 and
𝑅𝑠,𝑟𝑜𝑡𝑜𝑟 are known parameters that have been measured and calculated (Eq. 3.6) to an operating
temperature, in this case, 75 ℃ from 20 ℃. 𝑘 is the excitation constant, assumed to be 1.1 [41]
and [40]. In Åbjøra, the excitation constant k was approximated based on eight provided
measurements of excitation loss under various operating conditions, see Appendix F. Thus, the
well-known least square method was used to determine a more accurate estimate of the
excitation constant k. The leas square method was used as follows:
2 2
k = min (𝑒𝑟𝑟𝑜𝑟𝑖 ) = (𝑃𝑅,𝑚𝑒𝑎𝑠,𝑗 − 𝑘𝑖 ∙ 𝐼𝑓𝑑,𝑗 ∙ 𝑅𝑠,𝑟𝑜𝑡𝑜𝑟 )
where the excitation constant 𝑘 is determined from a “random” value of 𝑘𝑖 which produce the
least error to the measurement 𝑃𝑅,𝑚𝑒𝑎𝑠,𝑗 . 𝑖 represents the number for each value of 𝑘𝑖 , 𝑃𝑅,𝑚𝑒𝑎𝑠,𝑗
is the measurement of the total rotor loss, including excitation loss, where 𝐼𝑓𝑑,𝑗 is the
representative field current at each measurement 𝑗. With the least square method, the difference
between estimated and measured excitation constant showed a maximum error of about 2.5 %,
with an average error of 0.2 %.

57
 5 Methods
5.5.1 Approximation of field current
In this report, the field current was approximated with a method that was based on a method
used by Bortoni [53] and combined with some assumption described in [54]. The method
chosen considers saturation effects and can produce results with high accuracy. Compared to
most other methods, the main benefit of this method is how saturation considered. Saturation
can be implemented from the OCC by using either polynomials or an equation to represent the
OCC. The last was the method of choice.
The method requires information about 𝑃𝑡 [Pu], 𝑄𝑡 [Pu], 𝑉𝑡 [Pu], and generator parameters 𝑋𝑑 ,
𝑋𝑞 , 𝑋𝑑′ , 𝑋𝑙 , 𝑅𝑎 , with air-gap line and OCC to calculate the field current using the Potier
reactance. In this report, the generator parameters and characteristics were based on a document
4
where these parameters were estimated. The parameters used for Åbjøra and Sundsbarm can
be found in Appendix F and Appendix G, respectfully. The document gives only an estimate
of the parameters, and actual values may deviate slightly from these estimates. In Åbjøra, only
𝑋𝑞 , 𝑋𝑑′ , 𝑋𝑙 were used from the document, as actual measurements were available for the other
parameters and characteristics. It should be mentioned that the transient reactance 𝑋𝑑′ was given
as 0.29 Pu for Åbjøra but was adjusted to 0.15 Pu in this report to achieve similar estimations
of OCC and field current as what was measured. For Sundsbarm, all parameters and
characteristics were based on the document and could therefore be less accurate.
The following procedure has been used for field current approximation and is based on the
work by Bortoni [53] and Kundur [50]:
• Calculating the armature current 𝐼 and phase angle 𝜑:

√𝑃𝑡2 + 𝑄𝑡2
𝐼𝑡 = 5.10
𝑉𝑡
𝑄𝑡
𝜑 = tan−1 ( ) 5.11
𝑃𝑡
• The internal rotor angle, or load angle 𝛿𝑖 can then be calculated [53],[50]:
𝑋𝑞 𝐼𝑡 𝑐𝑜𝑠𝜑 − 𝑅𝑎 𝐼𝑡 𝑠𝑖𝑛𝜑
𝛿𝑖 = tan−1 ( ) 5.12
𝑉𝑡 + 𝑅𝑎 𝐼𝑡 𝑐𝑜𝑠𝜑 + 𝑋𝑞 𝐼𝑡 𝑠𝑖𝑛𝜑

where 𝑋𝑞 is the synchronous reactance on the q-axis, and 𝑅𝑎 is the armature winding
resistance.

• Calculating the internal excitation voltage:

𝐸𝑔 = 𝑉𝑡 𝑐𝑜𝑠𝛿𝑖 + 𝑅𝑎 𝐼𝑡 𝑐𝑜𝑠(𝜑 + 𝛿𝑖 ) + 𝑋𝑑 𝐼𝑡 𝑠𝑖𝑛(𝜑 + 𝛿𝑖 ) 5.13

4
The document is an excel sheet created by Prof. Hegglid, which estimates generator parameters based on factory
data.

58
 5 Methods
• In unsaturated operation or at open circuit, the field current 𝐼𝐹𝑈 have a linear
characteristic, i.e., a constant relationship between the excitation voltage and field
current. This relationship is usually known as air-gap line characteristic of the machine,
yielding:
𝐸𝑔
𝐼𝐹𝑈 = 5.14
𝑏𝑣
where 𝐸 [Pu] is the excitation voltage, 𝐼𝐹𝑈 is the unsaturated field current and 𝑏𝑣 is the
relationship between excitation voltage and field current in the air-gap, and determined
by reading off the air-gap line.
• Under magnetic saturation, the machine requires a higher field current to induce the
same voltage relative to the unsaturated operation. Thus, the so-called Potier reactance
is used as used to represent the saturation effects of the machine. From [54] [55] the
Potier reactance can be calculated with:
𝑋𝑝 = 𝑋𝑙 + 0.63(𝑋𝑑′ − 𝑋𝑙 ) 5.15
and if 𝑋𝑙 is not available, then Potier reactance for salient pole machines will be [54]
𝑋𝑝 = 0.7𝑋𝑑′ 5.16
where 𝑋𝑙 is the leakage reactance and 𝑋𝑑′ is the synchronous reactance.

• From [53], the Potier voltage will then be:

𝐸𝑝 = 𝑉𝑡 𝑐𝑜𝑠𝛿𝑝 + 𝑅𝑎 𝐼𝑡 𝑐𝑜𝑠(𝜑 + 𝛿𝑝 ) + 𝑋𝑝 𝐼𝑡 𝑠𝑖𝑛(𝜑 + 𝛿𝑝 ) 5.17

where
𝑋𝑝 𝐼𝑡 𝑐𝑜𝑠𝜑 − 𝑅𝑎 𝐼𝑡 𝑠𝑖𝑛𝜑
𝛿𝑝 = tan−1 ( ) 5.18
𝑉𝑡 + 𝑅𝑎 𝐼𝑡 𝑐𝑜𝑠𝜑 + 𝑋𝑝 𝐼𝑡 𝑠𝑖𝑛𝜑

• One could describe the curve with a high order polynomial equation by knowing the
OCC, as used in [53] and [54]. Another method to describe the OCC was mention in
[54], where it assumed most polynomial factors to be zero and was expressed as:
𝐼𝑂𝐶𝐶 = (𝐸𝑝 + 𝐶𝑛 𝐸𝑝𝑛 ) ∙ 𝑘 5.19

where 𝐼𝑂𝐶𝐶 is the field current given by the approximated OCC, k is set to nominal field
current (𝐼𝐹,𝑛 ), used to express the field current in [A] and not [Pu], 𝐸𝑝 is the Potier
voltage, 𝑛 is the most significant polynomial order, often set to 5, 7 or 9 [54]. The 𝐶𝑛
is a constant used to fit the approximated OCC curve, and in this report, it was
determined by trial-and-error method.

• The additional field current caused by saturation will then become:

𝐸𝑝
𝐼𝐹𝑆 = 𝐼𝑂𝐶𝐶 − 5.20
𝑏𝑣

59
 5 Methods

• The total field current will be the sum of unsaturated field current and the saturated
field current, yielding:
𝐼𝑓𝑑 = 𝐼𝐹𝑈 + 𝐼𝐹𝑆 5.21

A calculation example of this method and procedure can be seen in Appendix E.

5.5.2 Capability diagram

The capability diagram has been frequently used in this report to depict the boundaries of the
generators. The procedure used in this report is based on the work by Pejovski [49], which
shows how the capability diagram could be drawn and implemented in MATLAB. All
capability diagrams depicted in the report refers to the nominal voltage. The procedure
implemented is as follows:
• The armature current limit is drawn as a semicircle centred at the origin (0,0) with a
radius equal to the machine's MVA rating (𝑆𝑛 ), as illustrated in Figure 5.6. The nominal
active power to the generator is depicted as a horizontal line marked as (𝑃𝑚𝑎𝑥 ).

Figure 5.6: Armature current limit and maximum active power (𝑃𝑚𝑎𝑥 ) where 𝑃𝑚𝑎𝑥 = 𝑆𝑛 ∙ cos 𝜑𝑛 .The figure is
based on [49].

• Field current limit is determined by the relationship between active power (P) and
reactive power (Q) under a constant voltage (nominal) and excitation voltage (nominal),
i.e., constant field current. By assuming zero resistance, the equations for determining
the field current yields [49]:

60
 5 Methods
𝐸𝑉 𝑉 2 𝑋𝑑 − 𝑋𝑞 5.22
𝑃= 𝑠𝑖𝑛𝛿 + ∙ sin 2𝛿
𝑋𝑑 2 𝑋𝑑 ∙ 𝑋𝑞

𝐸𝑉 2
𝑋𝑑 − 𝑋𝑞 2
𝑉2
𝑄= 𝑐𝑜𝑠𝛿 + 𝑉 ∙ cos 𝛿 − 5.23
𝑋𝑑 𝑋𝑑 ∙ 𝑋𝑞 𝑋𝑞

𝐸 = 𝑉𝑐𝑜𝑠𝛿 + 𝑋𝑑 𝐼𝑠𝑖𝑛(𝜑 + 𝛿) 5.24

𝐼𝑐𝑜𝑠𝜑
𝛿 = arctan ( ) 5.25
𝑉
𝑋𝑞 + 𝐼𝑠𝑖𝑛𝜑

where 𝐸 is the generator’s excitation voltage

𝑉 is the generator’s phase voltage
𝐼 is the generator’s phase current
𝑋𝑑 and 𝑋𝑞 are the synchronous reactance’s, given as construction parameters
𝛿 is the load angle
𝜑 is the phase angle between voltage and current
From equation 5.23, one can see the centre of field current circle on the abscissa point
𝑉2 𝑋𝑑 −𝑋𝑞 𝑉2 𝑉2
( 2𝑛 ∙ − 𝑋 , 0) or reduced to (𝑋 , 0) as depicted in Figure 5.7. The field current
𝑋𝑑 ∙𝑋𝑞 𝑞 𝑑
limit can be drawn with equation (5.22) and (5.23) as coordinates, calculated under
nominal excitation voltage 𝐸 = 𝐸𝑛 at 𝛿 = 𝛿𝑛 . The filed current limit circle can then be
drawn varying the load angle (𝛿) in the range from 0 to 𝛿𝑛 , see Figure 5.7.

Figure 5.7: Field current limit. The figure is based on [49].

61
 5 Methods

• The end region heating limit or minimum field current limit can be determined with
equations equation 5.22 and 5.23, similar to the field current limit. Instead of using the
nominal excitation voltage as used in field current limit (maximum), the excitation
voltage is set to the minimum excitation voltage, given by:
𝐸𝑚𝑖𝑛 = 𝑘 ∙ 𝐸𝑚𝑎𝑥 5.26
where k-factor is 𝑘 = 0 − 0.3 [49]. In this report, the k-factor have been set to 0.3, see
Figure 5.8.
• The theoretical maximum power angle is 𝛿𝑚𝑎𝑥 = 90°, and will be located at the point
where active power in equation 5.22 peaks. The theoretical stability limit can therefore
by found by [49] [56]:
𝑑𝑃 𝐸𝑉 𝑋𝑑 − 𝑋𝑞
= 𝑐𝑜𝑠𝛿 + 𝑉 2 ∙ cos(2𝛿) = 0 5.27
𝑑𝛿 𝑋𝑑 𝑋𝑑 ∙ 𝑋𝑞

where 𝐸 ∈ (0,1).
Solving 5.27 for each excitation voltage E gives the point where active power P peaks
on the so-called Pascal curve, illustrated in Figure 5.7. The point where active power is
at maximum is called the intersection angle 𝛿, and can be used to draw the TLS from
equation 5.22 and 5.23.
Another approach to finding the theoretical stability limit is to use equation 5.28 and
5.29, derived from 5.27 [49] [56]:

√2√𝑘√𝑘 2 + 8𝑑 2 + 4𝑑 2 − 𝑘 2 ∙ (√𝑘 2 + 8𝑑 2 + 3𝑘) 5.28

𝑃𝑇𝑆𝐿 =
16𝑑
𝑘√𝑘 2 + 8𝑑 2 + 4𝑑 2 + 8𝑑𝑄0 − 𝑘 2 5.29
𝑄𝑇𝑆𝐿 =
8𝑑
𝐸𝑉 𝑑 1 𝑋𝑑 − 𝑋𝑞 2 𝑉2
𝑘= ; 𝑎= = ∙ 𝑉 ; 𝑄0 = − 5.30
𝑋𝑑 2 2 𝑋𝑑 ∙ 𝑋𝑞 𝑋𝑞

where 𝐸 ∈ (0,1).

• The practical stability limit will have a reduced value to the theoretical stability limit.
In this report, the practical stability curve is created by first drawing several excitation
curves (Figure 5.7) with different values of (𝐶), expressed in 5.31. The value of (C) is
determined by the practical stability curve, which intersects the crossing point
between armature and active power limit.

62
 5 Methods

𝑘√𝑘 2 + 8𝑑2 + 4𝑑 2 + 8𝑑𝑄0 − 𝑘 2

𝑄𝑃𝑆𝐿 = +𝐶 5.31
8𝑑

√2√𝑘√𝑘 2 + 8𝑑2 + 4𝑑 2 − 𝑘 2 ∙ (√𝑘 2 + 8𝑑 2 + 3𝑘) 5.32

𝑃𝑃𝑆𝐿 =
16𝑑

Figure 5.8: Capability diagram including practical stability limit, theoretical stability limit, end region heating
limit (k=0.3), field current limit (k=1) and armature current limit. The dotted curves represent the field current
under different excitation voltages (k=0.1-1). The figure is based on [49]

5.6 Transformer model

The transformer model is a relatively simple model consisting of a single equation presented
earlier in section 3.4, which only depends on the loading [MVA]. The simple model was chosen
as both no-load losses (𝑃0 ) and load losses (𝑃𝑘 ) are excessively used parameters on power
transformers nameplates. In addition, transformers usually have efficiencies in the range of
99.7 % under nominal rating, meaning an advanced model with higher accuracy will result in
neglectable difference when compared to the entire HPU efficiency.
For this report, the operational data was measured from the generator terminals and not from
the transformer terminals, which is assumed in the HPU model as it was initially developed
with regards to being used in combination with other simulation tools in the future. Instead of
changing the HPU model, the operational data was assumed to be measured at the high voltage

63
 5 Methods
terminals of the transformer, meaning the transformer losses needs first to be subtracted. The
operational data seen from the transformer side will then become:
𝑆𝑙𝑜𝑎𝑑 2
𝑃𝑡𝑟𝑎𝑛,𝑜𝑢𝑡 = 𝑃𝑡𝑟𝑎𝑛,𝑖𝑛 − 𝑃𝑙𝑜𝑠𝑠,𝑡𝑟𝑎𝑛 = 𝑃𝑔𝑒𝑛(𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙 𝑑𝑎𝑡𝑎) − 𝑃0 + 𝑃𝑘 ( )
𝑆𝑛

where 𝑃𝑡𝑟𝑎𝑛,𝑜𝑢𝑡 is the data used in the model, 𝑃𝑡𝑟𝑎𝑛,𝑖𝑛 = 𝑃𝑔𝑒𝑛(𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙 𝑑𝑎𝑡𝑎) is the original data.

5.7 Evaluation of hydropower generation

In this report, the efficiency and energy loss is analysed with different methods, where methods
like weighted average efficiency (WAE) and expected average efficiency (EAE) are used. The
explanation of WAE and EAE will be described in section 5.7.1 and 5.7.2, respectfully.

5.7.1 Weighted average efficiency

The weighted average efficiency (WAE) is a method that has been used to evaluate the
performance of the operational regime of hydropower. The WAE estimates the average
efficiency that considers all loading conditions where each loading is weighted relative to the
operational time [8]. Studies related to WAE can be seen in [8] and [53], where both are
extensive work by Bortoni, who used the WAE method for evaluating the operational
performance of different generators.
The WAE produces a single value 𝜂𝑤 , a number considering all the loading conditions of the
HPU, where efficiencies are weighed based on their probability of occurrence, i.e., the
percentage of time the machine operates at a given load, illustrated in Figure 5.9 [8]. The
weights are expressed as:
∆𝑡𝑘
𝐴𝑘 = 5.33
𝑇

WAE is the summation of operational efficiencies with their respectful weights, yielding:
𝑛

𝜂𝑤 = ∑ 𝐴𝑘 𝜂𝑘 5.34
𝑘=1

given that
𝑛

∑ 𝐴𝑘 = 1 5.35
𝑘=1

where ∆𝑡𝑘 is the total duration of each loading point 𝜂𝑘 , 𝑇 is the total duration time (typically
a day, month or year), 𝜂𝑤 is the weighted average efficiency, and 𝑛 is the number of loading
points. If all operations provided were accounted for; T = 4 ∙ 8256 hours (22.01.2020 –
31.12.2020). However, all operations with an active power of zero were neglected, resulting in

64
 5 Methods
a value of T = 28452 for Åbjøra and T = 25089 for Sundsbarm. In this report, the duration ∆𝑡𝑘
was determined based on the number of operations at a given efficiency.

Figure 5.9: The time-depending efficiency trend for WAE calculation [8].

5.7.2 Expected average efficiency

Expected average efficiency (EAE) is a new term proposed in this thesis, which is a method
that estimates the expected WAE for a reduced operating regime or operating points that are
disregarded. With the EAE, one can determine which operating regions limit the WAE the
most and what the WAE would be if certain regions are removed. The EAE is defined as the
WAE within an expected operational area, determined from the least expected efficiency, as
illustrated in Figure 5.10. The equation for EAE is in principle equal to the WAE where the
summation area of (5.34) is replaced by 𝜂𝑚𝑎𝑥 and 𝜂𝑚𝑖𝑛 , yielding:
𝜂𝑚𝑎𝑥

𝜂𝐸𝐴𝐸 = ∑ 𝐴𝑘 𝜂𝑘
𝑘=𝜂𝑚𝑖𝑛

65
 5 Methods

Figure 5.10: Illustration of EAE, where EAE is the WAE is estimated for all operations within the green area,
here 91 %. [own work]

66
 6 Power loss and efficiency analysis

6 Power loss and efficiency analysis

In this chapter, the energy loss and efficiency of the Åbjøra and Sundsbarm HPU will be
analysed and structured as shown in Figure 6.1. First, the losses in each component are
compared under nominal operation to give an overview of the magnitude and distribution of
losses. Then an in-depth loss analysis of each component where the effect of external influences
and operations are analysed to determine the BEP for each component and for the entire HPU.
Finally, a demonstration of the actual production of both HPUs will be analysed, where it will
be shown the operational regimes with the respective losses. In addition, a comparison between
both of the HPUs and their BEP will be provided to identify possible optimal and suboptimal
operating regimes in terms of energy loss. A summary of the analysis results can be found in
Table 6.1.

Figure 6.1: Chapter structure.

Table 6.1: Summary of operational statistics of Åbjøra and Sundsbarm. Operating time determined from
nominal turbine power and total energy production. Results from the period (22.01.20 – 31.12.20).

Power WAE Average Average Energy Energy BEP BEP

plant [%] P [MW] Q production loss [%] location
[MVAr] [GWh] [GWh] (Q,P) [Pu]
Åbjøra 91.6 79.31 1.03 579.6 53.1 92.25 (-0.22, 0.74)
Sundsbarm 91.87 94.4 3.54 602.8 52.2 92.14 (-0.30, 0.82)

67
 6 Power loss and efficiency analysis
6.1 Loss distribution
Analysis of the loss distribution is practised obtaining a better perspective on the size of the
accumulated losses and the efficiencies in each component. The majority of all losses are in
most cases determined by the active power [MW], and one could therefore depict the
distribution of losses from the relationship between efficiency and the active power, as depicted
in Figure 6.2. The figure shows the relationship between the power output (load) and the
efficiency of Åbjøra under nominal generator ratings. For Sundsbarm, a representative figure
can be seen in Appendix J. Figure 6.2 indicates a maximum HPU efficiency of 92.1 % achieved
around 78 MW. From the figure, it is evident that the turbine is the dominating factor, in
particular under low power levels as the turbine efficiency (max 94.5 %) is considerably lower
than the other components. However, under higher power levels, one can expect waterway
losses to become severe. The waterway efficiency in Åbjøra has shown to be about 1 % lower
than Sundsbarm, under nominal rating, even though the flow rate is approximately 24 [𝑚/𝑠]
for both HPUs. In both HPU’s, one could expect a transformer and generator efficiency above
99.5 % and 97 %, respectfully, which is typical for such large electrical machines.

Figure 6.2: Efficiency curves in relation to power output (measured at transformer output) in Åbjøra. Calculated
with: (𝑐𝑜𝑠𝜑𝑖𝑛𝑑 = 0.9), 𝑉 = 11 𝑘𝑉 and 𝐺𝑟𝑜𝑠𝑠 ℎ𝑒𝑎𝑑 = 442𝑚.

68
 6 Power loss and efficiency analysis
In Figure 6.3, it is shown how the losses are distributed under nominal generator rating; Åbjøra
(93 MW) and Sundsbarm (100 MW). Under nominal generator rating, the total power loss is
8.61 MW in Åbjøra, whereas the distribution indicates about 60% of all losses are related to
the turbine. In Sundsbarm, the total power loss is 9.45 MW, whereas 69 % is produced by the
turbine alone. In the figure, there is a clear difference in waterway losses between Åbjøra and
Sundsbarm, which is caused by the high head loss coefficient (K) in Åbjøra. The head loss
coefficient is about 1.83 times greater in Åbjøra, and since the nominal discharge (𝑄𝑓𝑙𝑜𝑤 ) is
about equal, one could expect an equally representative change in losses, as reflected in Figure
6.3. Under reduced load (0.3 Pu), the distribution of losses become significantly changed, like
the losses related to the turbine, which account as much for as 82 % (Åbjøra) and 80 %
(Sundsbarm) of the total losses.

Transformer 3 %
Generator 13 %
Waterway 24 %

Turbine 60 %

a)
Transformer
Waterway
4% Generator
13 %
14 %

Turbine
69 %

b)
Figure 6.3: Loss distribution under nominal generator operating point of a) Åbjøra and b) Sundsbarm.

69
 6 Power loss and efficiency analysis
6.2 Efficiency characteristics
The dominating factor regarding efficiency in an HPU is usually the active power, but external
influences like change in head, temperature or other factors may affect the efficiency. Thus, in
this section, the efficiency of the different components will be analysed with respect to external
influences.
In general, one could expect power losses in the waterway that are proportional to the flow rate
cubed, see equation 5.6. Under constant power output, there is a linear relationship between
head and flow rate, i.e., the power loss related to change in the head will be significant, as
illustrated in Figure 6.4.

Figure 6.4: Power loss in the waterway (Åbjøra), showing the relationship between power loss [𝑀𝑊] and flow
rate [𝑚3 /𝑠] relative to gross head [𝑚] under constant power (95 MW).

70
 6 Power loss and efficiency analysis
In Figure 6.5, it is shown an iso-efficiency contour map describing the turbine efficiency
(Åbjøra) under varying power production [𝑀𝑊] and gross head [𝑚]. A vital aspect of turbine
efficiency is the flow velocity. The flow velocity is, of course, important and related to friction
losses, but most importantly, the runner speed is design to operate at a fixed speed, and the
flow velocity is proportional to the square root of the head. Thus, the turbine would have a
specific head where the BEP is located, which can be seen from the hill-shaped map in Figure
6.5. Although the turbine has a specific optimum head, the difference in optimum efficiency is
only 0.02 % (94.56 – 94.54 %) under the specific range in the head.

Figure 6.5: Iso-efficiency contour map of the Francis turbine used in Åbjøra.

Due to the dominating effect of the turbine on the entire HPU, the hill-shaped structure depicted
in Figure 6.5 will also be present and similar to complete HPU efficiency, as seen in Figure
6.6. Figure 6.6 is an iso-efficiency contour map of the complete HPU in Åbjøra and Sundsbarm
and reveals a maximum HPU efficiency of 92.12 % and 91.98 %, occurring at a production of
77.7 MW and 95.0 MW with a gross head of 442.6 m and >490 m, respectfully.

71
 6 Power loss and efficiency analysis

a)

b)
Figure 6.6: Iso-efficiency contour map of the entire HPU, showing the relationship between efficiency, active
and head in a) Åbjøra and b) Sundsbarm. Reactive power and terminal voltage are set to nominal values.
In the generator and transformer, there are load and no-load losses. The relationship between
those losses produces an efficiency curve with a maximum efficiency at the point where load
and no-load losses intersect. The transformer model is simplified and assumed to only depend
on the absolute value [MVA] of the production. The transformer used in Åbjøra and Sundsbarm
has a maximum efficiency of 99.79 % (50 MVA) and 99.65 % (85 MVA), respectively, where
the results can be seen in Appendix I and Appendix J.

72
 6 Power loss and efficiency analysis
The optimum voltage has been found to be located at the nominal voltage for both HPUs (see
Appendix I and Appendix J), which can be expected as the terminal voltage is usually a “free
variable” during the designing process and selected based on the maximum efficiency point. In
the following simulations, the generator voltage will be set to nominal voltage 11 kV (Åbjøra)
and 15 kV (Sundsbarm).
The reactive power in a synchronous generator is fully controllable, which could be regulated
independently of the active power and therefore have a maximum operating point relative to
the excitation losses in addition to the active power. The iso-efficiency contour map of the
generator efficiency of Åbjøra is shown in Figure 6.7, where the efficiency is depicted relative
to the active and reactive power, overlayed with the representative capability diagram. The
corresponding figure for Sundsbarm can be found in Appendix J. In Figure 6.7, one can see the
generator having a BEP of 99.08 % achieved with an active power production of 1.4 Pu and
reactive power of -0.22 Pu, or a PF of 0.988. The best operating point is a theoretical optimum
and is not a practical optimum as the generator is limited by the capability diagram shown in
red. Within the capability diagram, the Åbjøra generator is limited to maximum efficiency of
99.0 %, with an active power production of >0.93 Pu and a reactive power of about -0.2 Pu.
For the generator efficiency, the influence of reactive power is relatively small compared to
active power production. The reactive power can only influence the efficiency to about +/-
0.2% under the nominal active power.

Figure 6.7: Efficiency mapping of the synchronous generator used in Åbjøra.

The BEP of the entire HPU has been obtained by plotting the active and reactive power
combined with the optimum head given by the results from Figure 6.6. The results for Åbjøra
and Sundsbarm can be seen in Figure 6.8. In Åbjøra, the BEP is 92.25 % achieved with a gross
head of 442.6 m, active power production of 0.74 Pu (76.2 MW) and reactive power of -0.22
73
 6 Power loss and efficiency analysis
Pu (22.7 MVAr). In Sundsbarm, the BEP is 92.14 % achieved with a gross head of 490 m,
active power production of 0.82 Pu (96.8 MW) and reactive power of -0.30 Pu (35.4 MVAr).
It should be noted that the location of the BEP is different. The BEP of Åbjøra is located further
down concerning active power, compared to Sundsbarm, which can be beneficial for HPUs
that frequently operates with low production, but certainly a disadvantage for HPUs operating
at rated power. In Sundsbarm, the BEP is located close to the rated power and does achieve
higher efficiency for all operations above 0.8 Pu compared to Åbjøra.
Figure 6.7 and Figure 6.8 show that reactive power has a minor influence on efficiency,
particularly compared to the total HPU efficiency, depicted in Figure 6.8. For both HPUs, on
can expect about 0.2 – 0.3 % deviation in the HPU efficiency caused by the reactive power.

a)

74
 6 Power loss and efficiency analysis
b)
Figure 6.8: Iso-efficiency contour map of the entire HPU, showing the relationship between efficiency, active
and reactive power under the optimal gross head, in a) Åbjøra and b) Sundsbarm.

To summarise, there has been analysed three variables (headwater, active and reactive) that
influence the efficiency. The results indicate a minor influence on the efficiency related to head
and reactive power, as the active power dominates the variations.
All components will have their own local BEP efficiency and are usually different from the
entire HPU. It is the sum of all efficiencies that result in the BEP for the HPU as a whole, which
is clearly shown in Table 6.2 and Table 6.3. The tables are a summary of the results of Åbjøra
and Sundsbarm.

Table 6.2: Optimum operation points at Åbjøra

Component Optimum load Optimum load Optimum head Maximum efficiency

(𝑃) [𝑀𝑊] (𝑄)[𝑀𝑉𝐴𝑟] [𝑚] [%]

Waterway 0 − High as possible ~ 100

Turbine 84 − 440 94.56

Generator 144 −22.7 − 99.08

Transformer 50 [𝑀𝑉𝐴] − 99.79

HPU (total) 76.2 −22.7 442.6 92.25

Table 6.3: Optimum operation points at Sundsbarm

Component Optimum load Optimum load Optimum head Maximum efficiency

[MW] [MVAr] [m] [%]

Waterway 0 − High as possible ~ 100

Turbine 100 − 490 93.91

Generator 212.4 −35.4 − 99.15

Transformer 85 [𝑀𝑉𝐴] − 99.65

HPU (total) 96.8 −35.4 490 92.14

75
 6 Power loss and efficiency analysis
6.3 Analysis of operational regime
The annual (2020) production regime of Åbjøra and Sundsbarm is shown in Figure 6.9 and
Figure 6.10, respectfully. The figures show both the active and reactive power production
measured at the generator terminals.
In Figure 6.9 and Figure 6.10, there are areas with high variations that could be a sign of
frequency restoration reserve (FRR) control, where the generator is regulated to balance the
grid. Areas with large spikes which shows zero production indicate the generator is situated in
an on/off situation. In addition, areas with large variation in production could indicate the
generator is turning on/off for a short period (under 1 hour), resulting in artificially low
production. As discussed earlier in section 5.2, the data was given as average values from
periods of 1 hour, making it difficult to distinguish between an on/off situation and a reduced
operation. Thus, the results from the average efficiency may be affected and deviate from the
actual operations.

Figure 6.9: Production regime of Åbjøra, showing both active and reactive production.

76
 6 Power loss and efficiency analysis

Figure 6.10: Production regime of Sundsbarm, showing both active and reactive production.
In order to have a better understanding of the operation, a so-called scatter diagram (left figure
in Figure 6.11) has been used to show the distribution of all operational data points (active and
reactive power). The figure to the left in Figure 6.11 depicts all operational data in relation to
the representative capability diagram and the efficiency. A single operational point is marked
as a blue dot in the scatter diagram, and areas with brighter colours represent a higher
operational density. The operational density can also be seen in the figure to the right, which
shows the percentage of all operations divided into given areas.
From Figure 6.11, there is a significant difference between Åbjøra and Sundsbarm when it
comes to the distribution. Åbjøra has a larger number of operations at the lower end and more
evenly distributed compared to Sundsbarm. Sundsbarm operates about 56 % of all operations
in proximity to the maximum turbine rating, and only 20 % is neither close to maximum rating
nor in the off state, resulting in an average operating point located at 0.8 Pu (active power) and
0.03 Pu (reactive power). For Åbjøra, the distribution of operations close to turbine rating is
40 %, whereas 46 % are found below the turbine rating, resulting in an average operating point
located at 0.79 Pu (active power) and 0.02 Pu (reactive power). Although the distribution of
operations differs from one another, the average operating point is about equal in terms of per
unit. The reason is due to the high turbine rating in Åbjøra, which is partly due to a high nominal
active power in the Åbjøra generator with a PF of 0.9, compared to 0.85 for Sundsbarm.

77
 6 Power loss and efficiency analysis

a)

b)
Figure 6.11: Distribution of operational data for a) Åbjøra and b) Sundsbarm. The dotted horizontal line in the
left figure represents the generator's nominal rating (active power). The figure to the right shows the rough
distribution of all operating points.
The estimated WAE for Åbjøra is 91.6 %, and for Sundsbarm, it is 91.9 %, as shown in Figure
6.12. Figure 6.12 presents the operational efficiencies as cumulative probability and the
expected average efficiency (EAE). The high WAE in Sundsbarm is clearly reflected in Figure
6.12, where one can see about 84 % of all operations are above 92 % efficiency, compared to
Åbjøra with only 20 % of all the operations. About 50 % of all operations in Åbjøra are found
in the region around 91.4 – 91.6 % efficiency, which is mainly represented by the production
under maximum turbine rating and a large area with low production, seen in Figure 6.11.
The EAE depicted in Figure 6.12 is the new term proposed in this report and shows how the
efficiency may be improved if one eliminates certain operations. For Åbjøra, one can see a
close to linear EAE characteristic and shows that one must remove about the 20 % operations
with the lowest efficiency to achieve only a 0.1% gain in WAE. For Sundsbarm, the EAE
characteristic shows that by eliminating the 10 % operations with the lowest efficiency, one
78
 6 Power loss and efficiency analysis
may achieve a 0.15 % gain in WAE. Eliminating the 10 % operations with the lowest efficiency
correlates to all operations below about 0.7 Pu (active power) for Sundsbarm, seen from Figure
6.11. One should notice that EAE does not include operations in off state, which is included in
Figure 6.11.

Figure 6.12: Cumulative probability of all operational efficiencies for both Åbjøra (blue) and Sundsbarm (red).
The EAE is marked as large, dotted lines. Operations situated in off-state are not considered in the figure.
The smaller WAE of Åbjøra is reflected in Figure 6.11, where one can clearly see Åbjøra being
operated far from the BEP, unlike the Sundsbarm, which have the BEP close to where most of
the operations are originated. It has been speculated whether the WAE could have been
improved if the efficiency curve of the turbine had shifted to a higher or lower active power
level with equal characteristics, e.g., in the case of a new turbine. It is a purely hypothetical
scenario, where the efficiency curve of Åbjøra and Sundsbarm turbine is shifted by multiplying
the active power with a factor, e.g., 0.95, 1.0 and 1.05, depicted in Figure 6.13. It was
discovered that the WAE did not improve when the efficiency curve was shifted in neither of
the HPUs, but rather decreases significantly. Figure 6.13 shows that the original turbine
characteristic is optimum for the given operational regime. As the turbine characteristic is
shifted to lower production (0.95), the BEP increases, whereas the efficiency at the average
operating point decreases and results in a lower WAE. Thus, the highest WAE will be
determined by the turbine characteristic which has the highest efficiency at the average
production. One should have in mind that the location of the average production determines
which characteristic that results in the highest WAE but does not give any information about
the magnitude of the actual WAE (e.g., Figure 7.4).

79
 6 Power loss and efficiency analysis

Figure 6.13: A hypothetical scenario of Åbjøra showing the effect of shifting the turbine characteristic curve.
The turbine is given by interpolated measurements, which have been multiplied by 0.95, 1.0 (original) and 1.05
in order to move the turbine characteristic curve. The intersection point between HPU characteristic (blue) and
average production is marked with a red circle.

92.2
92.0 91.9 92.0 91.9 92.0
91.9 92.0
92.0 92.0
92.0 91.9
91.7 91.8
91.7 91.7
Efficiency [%]

91.8 91.7
91.6 91.6
91.6 91.6
91.6 91.5 91.5 91.5
91.4 91.3
91.4

91.2

91.0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Des
Month
Åbjøra Sundsbarm

Figure 6.14: WAE of Åbjøra and Sundsbarm.

Analysis of the monthly average efficiency (WAE) gives the results shown in Figure 6.14,
where Åbjøra is depicted in blue and Sundsbarm in orange. Åbjøra has the highest WAE in
January (91.89 %) and the lowest in April (91.35 %), whereas Sundsbarm have the highest
WAE in August (92.02 %) and the lowest in July (91.35 %).
The monthly operation with the highest and lowest WAE in Åbjøra and Sundsbarm is
illustrated in Figure 6.15 and Figure 6.16, respectfully. The figures are uses to illustrated

80
 6 Power loss and efficiency analysis
relationship between average production and BEP. Although the highest WAE can be realised
when the average production is close to the BEP, it does not imply that this is the case for all
situations. In the left figure in Figure 6.15, the average operating point is about identical to the
BEP, but most operations are either above or below the average point, resulting in a limited
WAE compared to a situation where all operations are located at BEP. When it comes to the
month (August) with the highest WAE in Sundsbarm, the difference between average
production and BEP is larger than in Åbjøra but still obtains a higher WAE as a high number
of operations is actually operating close to the BEP.

Figure 6.15: Production regime Åbjøra during January and April, with their representative WAE compared to
the BEP. The hours are referenced from the first measurement (22.01.2020).

Figure 6.16: Production regime Sundsbarm during August and July, with their representative WAE compared to
the BEP. The hours are referenced from the first measurement (22.01.2020).
In some situations, the WAE may not reflect the performance of the HPU as some months may
have long or short periods with no operation. Thus, analysing the energy production and energy
loss could give a better picture of the actual performance, as shown in Figure 6.17. In the figure,
one can see the total energy production (blue) and energy losses (orange) for each month, which
results in total energy consumption (water used). The annual (except the few days in January)

81
 6 Power loss and efficiency analysis
energy production of Åbjøra was 579.6 GWh with a total loss of 53.1 GWh, compared to
Sundsbarm, which produced 602.8 GWh with a total loss of 52.2 GWh.

Åbjøra
80.0
70.0 6.0 6.2 5.8
5.0 5.1 4.9 4.6
Energy [GWh]

60.0
50.0 4.0
3.6 3.5
40.0
2.8
30.0 64.2 67.1 62.9
55.1 55.5 54.7 51.0
20.0 1.5 38.5 39.0 44.6
30.3
10.0 16.7
0.0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Des
Date

Production Loss

a)

Sundsbarm
90.0
6.4 6.1 6.5
80.0 6.1 5.8
5.5
Energy [GWh]

70.0
60.0
50.0 4.0
40.0
73.9 70.5 2.9 3.0 74.9 66.5
30.0 2.1 63.4 2.3 2.1 69.5
20.0 46.6
23.8 32.3 24.8 32.2 24.4
10.0
0.0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Des
Date

Production Loss

b)
Figure 6.17: The monthly energy production and energy loss in a) Åbjøra and b) Sundsbarm.
A comparison between actual and optimal energy loss is seen in Figure 6.18, where the optimal
energy loss have been found by assuming both HPU’s were operated at BEP with equal
production. The results indicate a reduction in energy loss of 8.2 GWh and 5.4 GWh for Åbjøra
and Sundsbarm, respectfully. By considering the average system Elspot price (Nord Pool), one
can give a rough estimate of the possible savings related to energy losses. Due to the
extraordinary low prices in 2020, the prices will be based on 2019, where the average system
82
 6 Power loss and efficiency analysis
price was 383 NOK/MWh, resulting in an annual savings of 3.14 million NOK for Åbjøra and
2.07 million NOK for Sundsbarm.

7.0
6.2
6.0 5.8
6.0
5.1 5.0 5.2
5.0 4.9 4.9
5.0 4.6
Energy loss [GWh]

4.3 4.3 4.2

4.0 4.0
4.0 3.6 3.5 3.5
3.0 3.0
2.8
3.0 2.3
2.0 1.51.3

1.0

0.0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Des

Month
Actual Optimal

b)
7.0 6.4 6.5
6.1 6.1
5.8 5.9 5.8
6.0 5.5 5.5 5.5
5.2
Enegy loss [GWh]

5.0
5.0
4.0
4.0 3.7
2.9 3.0
3.0 2.5 2.5
2.11.9 2.3 2.11.9
1.9
2.0

1.0

0.0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Des

Month
Actual Optimal

a)
Figure 6.18: Energy loss a) Åbjøra and b) Sundsbarm. The optimal energy loss represents the energy loss that
would be obtained if operated at BEP under equal production as the actual data.

83
 7 Sensitivity analysis

7 Sensitivity analysis
In this chapter, a short sensitivity analysis will be presented to identify the WAE under new
operating regimes. In the future power grid, one may expect larger fluctuations in active and
reactive power demand, partly caused by the intermittency of wind and solar. A sensitivity
analysis could therefore be helpful for preparing for what to come in the future. It is difficult
to determine the actual future operating regime, so in this report, there will be focused on two
scenarios. Scenario 1) shall identify patterns in WAE by varying the active power production
while maintaining the maximum production and reactive power constant, and scenario 2) shall
vary the reactive power production while maintaining the active power constant, see Figure
7.1. Analysing the production of active and reactive power separately makes it possible to
examine the effect each of them has on the WAE.
The result from this analysis indicates that one could expect a significant reduction in WAE,
of which a 2 % reduction has been identified under severe changes in the active power regime.
On the other hand, the influence of reactive power on the WAE has shown to be minor, with a
maximum reduction of 0.02 % under large fluctuations of reactive power.

Figure 7.1: Illustration of the sensitivity analysis scenarios. The green area represents the original operational
regime, whereas the red and yellow area represents expansion and contraction of the operational regime,
respectfully.

84
 7 Sensitivity analysis
7.1 Scenario 1 - Variation of active power
In this scenario, the active power is varied by expanding and contracting the operation points
while maintaining the maximum active power, see Figure 7.2. The scenario is simulated ten
times, and each test is numbered as P1 to P10, where P10 represents full extension (𝑎𝑖 = 2)
and P1 represents full contraction (𝑎𝑖 = 0.2), where each test uses the following equation to
vary the active power:
𝑃𝑛𝑒𝑤,𝑖 = 𝑎𝑖 ∙ 𝑃𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙 − 𝑏𝑖
where 𝑎𝑖 is a factor that starts with a value of 0.2 and increases with increments of 0.2 up to 2
for each test 𝑖 and 𝑏𝑖 is a value used to maintain the maximum active power and is estimated
by:
𝑏𝑖 = 𝑎𝑖 ∙ 𝑃𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙,𝑚𝑎𝑥 − 𝑃𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙,𝑚𝑎𝑥
where 𝑃𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙,𝑚𝑎𝑥 is the maximum active power, 95 MW and 103 MW for Åbjøra and
Sundsbarm, respectfully. An overview of the numbers used for each test is shown in Table 7.1:
When performing this scenario, it is vital that all operations at off-state (𝑃 = 0) are removed,
else the off-state operations are evaluated as operations.
Table 7.1: Overview of test values used

Åbjøra
Test nr. P1 P2 P3 P4 P5 P6 P7 P8 P9 P10
𝑎𝑖 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
𝑏𝑖 [MW] -76 -57 -38 -19 0 19 38 57 76 95.0
Sundsbarm
Test nr. P1 P2 P3 P4 P5 P6 P7 P8 P9 P10
𝑎𝑖 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
𝑏𝑖 [MW] -82 -62 -41 -21 0 21 41 62 82 103.0

85
 7 Sensitivity analysis

Figure 7.2: Scenario 1, variation of active power, where left figures represent operations of Åbjøra and right
represents Sundsbarm. Test P1 shows full contraction (𝑎𝑖 = 0.2) and test P10 shows full extension (𝑎𝑖 = 2).

86
 7 Sensitivity analysis

The test results indicate a change in WAE, as depicted in Figure 7.3. The figure shows a
relatively large reduction in average efficiency when the operating regime has a higher density
of low active power operations (P10). Between test P10 and original data (P5) for Åbjøra, the
results show a WAE reduction from 91.6 % to 89.7 % or a difference of 1.9 %. For Sundsbarm,
the WAE was reduced from 91.9 % to 90.9 % or a difference of 1 %. From the tests P1-P4, it
was shown an increase in WAE for both HPU’s, where Åbjøra increased the WAE by 0.19 %
between maximum (P3) and original operation (P5), and Sundsbarm had a maximum increase
between test P1 and P5 with an increase of 0.1 %. It should be noticed that Åbjøra have a more
prominent change than Sundsbarm due to the difference in the original distribution of
operational density. The function used to create each test influences the lowest power levels
more than the highest, as these are maintained about constant.

Figure 7.3: Estimated WAE for each test

In Figure 7.4 it is shown the correlation between average active power for each test, the WAE
for each test and the BEP. The figure indicates no significant correlation between the average
active power obtained for each test and the active power representing the BEP. Both HPUs
achieved a maximum WAE under a higher average active power, see Figure 7.4.

87
 7 Sensitivity analysis

a)

b)
Figure 7.4: Correlation between average active power, BEP and WAE obtained from each test for a) Åbjøra and
b) Sundsbarm

88
 7 Sensitivity analysis
7.2 Scenario 2 – Variation of reactive power
In this scenario, the magnitude of the reactive power will be changed, this by multiplying the
reactive component of all operational points with a constant to achieve a higher variation of
reactive power, illustrated in Figure 7.5.
Within this scenario, there will be 11 simulations numbered from 𝑄1 – 𝑄11, where 𝑄1 represents
the original operation, and for each simulation, the multiplied factor k will increase in
increments of 0.2. The change in reactive power can be expressed as:
𝑄𝑛𝑒𝑤,𝑖 = 𝑘𝑖 ∙ 𝑄𝑜𝑟𝑖𝑔𝑖𝑛𝑎𝑙
where 𝑘𝑖 is the multiplication factor for each test 𝑖 and ranges from value 1.0 to 3.0 with
increments of 0.2.

a)

b)
Figure 7.5: Increased variation of reactive power, where a) is the operation in Åbjøra and b) is Sundsbarm. Test
Q11 represents a variation of reactive power three times the original.

89
 7 Sensitivity analysis

From the simulations, the WAE only reduced by about 0.02% between test Q1 and Q11 for
both HPUs. The corresponding annual energy loss would increase by 0.15 and 0.16 GWh for
Åbjøra and Sundsbarm, respectfully. As discussed earlier (Section 6.2), the reactive power can
only change the efficiency to about 0.2 – 0.3 % under all loading conditions. In this test
scenario, the reactive power is amplified with a constant (1 - 3.0), meaning all reactive
components that are negative (absorbing) will become closer to the BEP (-0.22 and -0.3 Pu)
and improve efficiency and limit the changes in WAE.

91.595 91.87

91.59 91.865

91.585 91.86
WAE [%]

WAE [%]
91.58 91.855

91.575 91.85

91.57 91.845

91.565 91.84

91.56 91.835
Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Q10 Q11
Test nr.

Åbjøra Sundsbarm

Figure 7.6: WAE obtained from each test for both Åbjøra and Sundsbarm. The test nr. represents each
simulation where all reactive power measurements are multiplied by a constant ranging from 1.0 – 3.0.

90
 8 Discussion

8 Discussion
8.1 Data acquisition and filtering process
The operational data acquired contained information about the terminal voltage, active and
reactive power production to both generators. The data could not be used without modifications
as it had apparent errors or problems which needed to be sorted out. The main issues were:
• A large set of operations was represented with “NaN” instead of the actual value.
• All data was given as average values from 1-hour periods.
These problems influence the results in different ways, and one may expect the results to
deviate slightly from reality. Average values may not directly affect the energy production, but
the combination of “NaN” and average values made it difficult to distinguish between start/stop
regions. The method used to differentiate between start/stop regions was by analysing the
magnitude of the terminal voltage, indicated the possible duration of an operation point. Thus,
an algorithm was created to filter the data based on given assumptions. Both HPUs frequently
operated in FRR, i.e., the excessive use of start/stop operations and large fluctuations in active
or reactive power. The combination of considerable variation in the active power and
measurements from average values would result in some uncertainty to the estimated WAE
and other efficiency calculations.
An example would be two operations operated both above and below the BEP with equal
distance from the BEP, concerning efficiency. This scenario would result in an average value
of active power close to the BEP. However, the average efficiency of those operations would
be far from the BEP. Because of the efficiency and energy losses, average values may not be
suited for this type of analysis, where one could expect a production with large fluctuations
within the average period, here 1-hour.

8.2 Simulation model and assumptions

In this report, the headwater was set to nominal during the simulations due to a lack of
accessible data. The effect of possible head difference have been analysed and indicate a BEP
reduction of about 0.06 % and 0.1 % in Åbjøra and Sundsbarm respectfully, and should be
considered in future work.
The HPU model was initially designed to use measurements from the high voltage side of the
transformer. Instead, the provided data was given for the generator terminals. A quick solution
to this problem was to calculate the power loss of the transformer ahead and add the power loss
to original operational data (P) before implementing the data in the simulation model. This
procedure will not affect the energy loss calculated in this report. In cases where the data is
provided from the grid side, the transformer inductance shall be implemented to consider the
reactive power absorption of the transformer, which was not necessary as the reactive power
to the generator was already known.

91
 8 Discussion
It was discovered that the WAE obtained the same results as the general formula for average
value, i.e., 𝑥̅ = (∑ 𝑥)/𝑁. The reason is that all measurements have the same duration (15
minutes), resulting in weights (𝐴𝑘 ) which are either equal or directly proportional to the number
of operations. The WAE will differ from the general formula for average value if each duration
is different or disproportionate to the number of operations.

8.2.1 Turbine and waterway model

The data provided for the turbine contained estimates of the efficiency under various operating
conditions. The estimates were performed in 2011 for both HPUs, meaning the effect of turbine
fatigue might be present but have been neglected in this report. In the turbine and waterway
data, it was described that the spiral casing and draft tube was a part of the waterway. Therefore,
it is uncertain whether or not the turbine data consists of only the runner efficiency or if swirling
losses (helical vortices) in the draft tube are considered a part of the turbine or the draft tube.
This detail may influence the accuracy of the waterway and turbine model.
The turbine model uses interpolation to estimate the efficiency, which is often a preferred
method. However, interpolation methods are known for generating artificial oscillations in the
data. In some of the iso-contour maps, the oscillations become prominent due to few data points
related to the head.

8.2.2 Generator model

In the generator model, excitation losses are assumed to be a constant proportional to the losses
in the field windings. The proportional constant of excitation may not reflect the reality but
gives a valid approximation as the excitation losses are close to linear. The constant was set to
the default of 1.1, the recommended value proposed by Westgaard [41]. For Åbjøra, excitation
losses were set to 1.116, determined by the least square method from multiple excitation
measurements to improve the accuracy. The accuracy had a maximum error of 2.5 % and an
average error of 0.2 %, which will not contribute to any significant errors related to the entire
HPU efficiency.
The field current calculations used a set of parameters and characteristics, which was provided
from estimates and measurements. Thus, the simulation results from the generator may differ
for the reality to a certain degree. In Åbjøra, the field current was determined from both
estimated parameters and actual measurements, which means some adjustments to the
estimated values were required to achieve results close to the actual measurements. The
generator model of Åbjøra was compared to the measurements, and high accuracy was
confirmed. However, there is no evidence of the accuracy of the estimated field current in
Sundsbarm.
The generator model neglects all effects of temperatures as no temperature data was accessible.
The temperature setting of armature and field windings was set to a constant value of 75 ℃,
i.e., the winding losses might differ marginally, particularly during lower power levels where
the temperature is limited.

92
 8 Discussion
In the generator model, the voltage is only affecting the armature and field winding current.
The effect of voltage in the iron core is neglected, which can be considered a valid assumption
as the voltage fluctuations are negligible.
It has been assumed windage and friction losses to be constant, which is usually valid when
the frequency is constant as both windage and friction heavily depend on the machine's
rotational speed. In addition, the ventilation losses have been assumed constant and are
considered a valid assumption, as the cooling of Åbjøra and Sundsbarm is unregulated air-
cooled machines, i.e., fans mounted directly on the rotor.

8.2.3 Capability diagram

For this report, the capability diagrams are primarily used for illustrational purposes. The
armature and maximum field limits can be regarded as reasonable. In contrast, the end heating
limit and practical stability limit do not necessarily represent reality as these were based on
some assumptions.

8.3 Discussion of results

The results have shown that the waterway losses in Åbjøra are considerable, as waterway losses
accounted for 24 % of the total losses under nominal power, compared to only 13 % in
Sundsbarm. The high waterway losses found in Åbjøra have a significant impact on efficiency
but may also be the most expensive component of all to improve. The high waterway losses of
Åbjøra are an essential factor contributing to moving the BEP to a lower power level which is
not preferable if operating the HPU at higher power levels.
The BEP of Åbjøra (92.25 %) was found to be slightly higher than Sundsbarm (92.14 %),
which is primarily due to the high turbine efficiency in Åbjøra. However, Åbjøra had a WAE
of 91.6 %, which are lower than obtained in Sundsbarm (91.9 %), and is a result of having a
BEP located at low power levels combined with large fluctuations in operation. Most of the
large fluctuations seen in both HPUs are produced by what is recognized as FRR or mFRR
operating regime, which means the HPUs are automatically or manually controlled by the
national TSO (Statnett) to balance the power grid. During the summertime, Sundsbarm is
primarily used for grid balancing, which also results in a significant operational efficiency
reduction. In contrast, Åbjøra obtains a low operational efficiency during the winter, where the
production is high.

8.4 Discussion of sensitivity analysis

The sensitivity analysis was performed to analyse how the WAE changes relative to operational
patterns. The analysis was chosen to vary the active and reactive power based on the original
regime and indicated a marginal increase in loss related to greater fluctuations of reactive
power, assuming a proportional increase. If the generators operate more frequently in the
inducive region, one could expect an improvement to the WAE. However, even in extreme
scenarios, the influence of reactive power only influenced the WAE of about 0.02 % in both
HPUs.

93
 8 Discussion
On the other hand, the change in WAE due to increased active power variation resulted in about
0.7 % and 2.8% change in Sundsbarm and Åbjøra, respectfully. The equation used to vary the
active power did not affect all operating points equally, as operating points in proximity to the
maximum power was almost not affected. In this sensitivity analysis, the variations in active
power levels were increased, which is not realistic as both generators operate at a minimum
setpoint. It is more likely that the operational density increase at the lower power levels or
become more distributed. One could expect Sundsbarm to be most affected by the change as
the location of the BEP is more favourable for high power levels, which is the opposite situation
for Åbjøra.

94
 9 Conclusion

9 Conclusion
The future power grid may experience severe changes due to the “green shift” which utilises
intermittent energy sources like wind and solar power. With extensive use of intermittent
sources, the power grid may experience severe challenges regarding grid balance, and
hydropower units could be forced to change their operational regime to focus more on
balancing the grid. This regime shift is expected to influence the hydropower units' operational
efficiency, resulting in severe energy losses and lost revenue.
Therefore, this report will analyse the operational efficiency and energy losses in hydropower
units to prepare for future challenges. The analysis was performed as a part of a larger research
program collaborated by Skagerak Kraft and the University of South-Eastern Norway (USN).
During this report, production regimes, regulations, and general hydropower theory has been
introduced with a focus on energy production and related losses. In this report, a static
hydropower model has been developed in MATLAB. The model combines basic fluid dynamic
theory and electrical machine theory, allowing energy loss and efficiencies to be determined
from parameters given by the hydropower unit and operational data from the generator
(transformer). The model has been implemented in two hydropower units, namely Åbjøra and
Sundsbarm. The analysis of these hydropower units is presented in two study cases. The first
study case analyses and compares the individual components and the entire hydropower unit
followed up with an analysis of the operational regimes. The second study case is a short
sensitivity analysis where changes in the operational regimes are investigated for possible
future grid changes.
From the analysis of Åbjøra and Sundsbarm, a total best efficiency point (BEP) was found with
an efficiency of 92.25 % and 92.14 %, respectfully. The results show signs of frequent use of
FRR in both hydropower units, which strongly correlates with reduced operational efficiency
due to the large fluctuations of active power. Sundsbarm operates most efficiently under high
power levels with a weighted average efficiency (WAE) of 91.9 %, whereas Åbjøra operates
with a WAE of 91.6 %. From the report, it can be concluded that the impact of reactive power
on efficiency will be minor and does not contribute to any significant change to the efficiency.
On the other hand, operations with low active power should be limited due to the considerable
impact on efficiency, which could occur with increased use of grid balancing.

95
 10 Further work

10 Further work
To improve the accuracy of this analysis, one should use single measurements or average
values with short time periods. Variations in the head should be implemented to account for
the additional losses in the waterway and turbine. Additional losses related to start/stop and
idle situations could be implemented, which may be significant for hydropower units with
frequent start/stop operations. The generator in Sundsbarm should be implemented with more
accurate parameters, preferably from measurements. The effect of fatigue in turbines should
be investigated and implemented in the model. For the transformer, an improved model could
be constructed, which includes the internal impedance for reactive power absorption.
The HPU model is only designed for single generating units and should be expanded for multi
generating units as additional head loss calculations may be required. For the long run, the
HPU model should be improved with a higher level of software usability.

96
 References

References
[1] ‘Kraftproduksjon’, Energifaktanorge, Mar. 25, 2021. https://energifaktanorge.no/norsk-
energiforsyning/kraftforsyningen/ (accessed Apr. 18, 2021).
[2] A. Joswig and M. Baca, ‘Extended requirements on turbo-generators and solutions for
flexible load operation’, in 2016 XXII International Conference on Electrical Machines
(ICEM), Lausanne, Switzerland, Sep. 2016, pp. 2649–2654. doi:
10.1109/ICELMACH.2016.7732895.
[3] E. F. Bødal, ‘Coordination of Hydro and Wind Power in a Transmission Constrained Area
using SDDP’, Norwegian University of Science and Technology, 2016.
[4] ‘Hydroelectricity in the Nordic countries’, Nordics.info, Feb. 25, 2019.
https://nordics.info/show/artikel/hydroelectricity/ (accessed May 12, 2021).
[5] ‘Meld. St. 25 (2015–2016): Kraft til endring - Energipolitikken mot 2030’, Det Kongelige
Olje- og Energidepartement, Apr. 2016.
[6] ‘Egenskaper ved det norske kraftsystemet’, Norges vassdrags- og energidirektorat
(NVE), Aug. 26, 2020. https://www.nve.no/stromkunde/om-kraftmarkedet-og-det-
norske-kraftsystemet#01 (accessed Apr. 18, 2021).
[7] I. V. Sem et al., ‘Langsiktig Kraftmarkedsanalyse 2020-2040’, NVE (The Norwegian
Water Resources and Energy Directorate), 37/2020, Oct. 2020. Accessed: May 13, 2021.
[Online]. Available: http://publikasjoner.nve.no/rapport/2020/rapport2020_37.pdf
[8] E. C. Bortoni, R. T. Siniscalchi, S. Vaschetto, M. A. Darmani, and A. Cavagnino,
‘Efficiency Mapping and Weighted Average Efficiency for Large Hydrogenerators’,
IEEE Open J. Ind. Applicat., vol. 2, pp. 11–20, 2021, doi: 10.1109/OJIA.2020.3048989.
[9] A. Estanqueiro, ‘The Future Energy Mix Paradigm: How to Embed Large Amounts of
Wind Generation While Preserving the Robustness and Quality of the Power Systems?’,
in Wind Power, S. M, Ed. InTech, 2010, p. 23. doi: 10.5772/8359.
[10] S. Fereidoon and S. P. Fereidoon, Future of utilities - utilities of the future : How
technological innovations in distributed energy resources will reshape the electric power
sector, 1st ed. London, England: Academic Press, 2016.
[11] J. K. Noland, M. Leandro, A. Nysveen, and T. Oyvang, ‘Future Operational Regimes of
Bulk Power Generation in The Era of Global Energy Transition: Grid Codes, Challenges
and Open Issues’, in 2020 IEEE Power & Energy Society General Meeting (PESGM),
Montreal, QC, Canada, Aug. 2020, pp. 1–5. doi: 10.1109/PESGM41954.2020.9282001.
[12] THE EUROPEAN COMMISSION, ‘COMMISSION REGULATION (EU)
establishing a network code on requirements for grid connection of generators (NC-
RfG)’, p. 68, Apr. 2016.
[13] Statnett, ‘NVF 2020 Nasjonal veileder for funksjonskrav i kraftsystemet’, p. 240, Jul.
2020.
[14] Statnett, ‘Funksjonskrav i kraftsystemet 2012’, p. 116, 2012.
[15] ‘IEEE Std C50.13TM-2014, Standard for Cylindrical-Rotor 50 Hz and 60 Hz Synchronous
Generators Rated 10 MVA and Above’, p. 63, Mar. 2014.

97
 References
[16] K. Mayor, L. Montgomery, and K. Hattori, ‘Grid code impact on electrical machine
design’, in 2012 IEEE Power and Energy Society General Meeting, San Diego, CA, Jul.
2012, pp. 1–8. doi: 10.1109/PESGM.2012.6345292.
[17] T. Kallevik et al., ‘RME RAPPORT: Driften av kraftsystemet 2019’, Norges vassdrags-
og energidirektorat (NVE), 3/2020, 2019. [Online]. Available:
http://publikasjoner.nve.no/rme_rapport/2020/rme_rapport2020_03.pdf
[18] Statnett, ‘Fast frequency reserves FFR’, Statnett, Jan. 22, 2021.
https://www.statnett.no/for-aktorer-i-
kraftbransjen/systemansvaret/kraftmarkedet/reservemarkeder/ffr/ (accessed May 16,
2021).
[19] ‘Primærreserver - FCR’, Statnett, Nov. 09, 2018. https://www.statnett.no/for-aktorer-i-
kraftbransjen/systemansvaret/kraftmarkedet/reservemarkeder/primarreserver/ (accessed
May 12, 2021).
[20] E. J. Ruud, ‘Beslutningsstøtte med kontinuerlig virkningsgradsmåling i vannkraftverk’,
Norwegian University of Science and Technology (NTNU), 2017.
[21] ‘ÅRSBERETNING OG ÅRSREGNSKAP /2018’, Sira Kvina Kraftselskap, 2018.
Accessed: May 14, 2021. [Online]. Available:
https://www.sirakvina.no/getfile.php/132809-
1559636215/Dokumenter/%C3%85rsberetninger/Sira-Kvina_%C3%A5rsberetning-
2018.pdf
[22] ‘How hyperlocal hydro energy can empower communities’, Grist.org, Nov. 11, 2020.
https://grist.org/Array/how-hyperlocal-hydro-energy-can-empower-communities/
(accessed May 14, 2021).
[23] I. E. Idel’chik, ‘Handbook of Hydraulic Resistance’, Israel Program for Scientific
Translations, p. 526, 1960 1966.
[24] R. K. Bansal, A Textbook of Fluid Mechanics and Hydraulic Machines, 9th ed. New
Delhi: Laxmi Publications, 2005.
[25] M. B. Bishwakarma, ‘Computation of Head Losses in Hydropower Tunnels’, p. 16, Nov.
2012.
[26] O.-I. Lekang, Aquaculture Engineering, 2nd ed. Ås, Norway: John Wiley & Sons, Ltd,
2013.
[27] A. N. Kamarudzaman, ‘EAT 252: Fluid Mechanics Engineering: Chapter 6 - Minor Loss’,
School of Enviromental Engineering Universiti Malayssia Perlis. Accessed: May 16,
2021. [Online]. Available:
http://portal.unimap.edu.my/portal/page/portal30/Lecture%20Notes/School%20of%20E
nvironmental%20Engineering/Semester%201%20Sidang%20Akademik%2020172018/
RK01/EAT252%20Fluid%20Mechanics%20Engineering/EAT%20252%20CHAPTER
%206%20Minor%20loss%20portal.pdf
[28] A. Kjølle, HYDROPOWER IN NORWAY Mechanical Equipment. Trondheim, Norway:
Norwegian University of Science and Technology (NTNU), 2001.
[29] ‘An Ultimate Guide to Francis Turbine’, Linquip Technews, Dec. 13, 2020.
https://www.linquip.com/blog/what-is-francis-turbine/ (accessed May 15, 2021).

98
 References
[30] ‘Turbines’. http://www.eternoohydro.com/turbines/axial-flow-turbines.html (accessed
May 15, 2021).
[31] ‘Design of Pelton turbines’, Norwegian University of Science and Technology (NTNU),
2006. [Online]. Available:
http://www.ivt.ntnu.no/ept/fag/tep4195/innhold/Forelesninger/forelesninger%202006/5
%20-%20Pelton%20Turbine.pdf
[32] ‘Pelton turbines’, Voith. https://voith.com/hu-hu/turbines-generators/turbines/pelton-
turbines.html (accessed May 15, 2021).
[33] Z. Zhang, Pelton Turbines, 1st ed. 2016. Cham: Springer International Publishing :
Imprint: Springer, 2016. doi: 10.1007/978-3-319-31909-4.
[34] S. Alligné, ‘Forced and Self Oscillations of Hydraulic Systems Induced by Cavitation
Vortex Rope of Francis Turbines’, ÉCOLE POLYTECHNIQUE FÉDÉRALE DE
LAUSANNE, 2011.
[35] S. Mathew, ‘Pelton turbine - working and design aspect’, Lesics, Aug. 21, 2013.
https://lesics.com/pelton-turbine-wheel-hydraulic-turbine.html (accessed May 15, 2021).
[36] A. Alfredsson and K. A. Jacobsson, Elmaskiner och elektriska drivsystem, 3rd ed.
Stockholm,Sweeden: Liber, 2016.
[37] A. Petterteig, O. Mogstad, T. Henriksen, and Ø. Håland, ‘Tekniske retningslinjer for
tilknytning av produksjonsenheter, med maksimum aktiv effektproduksjon mindre enn 10
MW, til distribusjonsnettet’, Sintef, Trondheim, Norway, Technical report TR A6343.01,
Nov. 2006. [Online]. Available:
https://www.sintef.no/globalassets/project/distribution_2020/publikasjoner/tr_a6343.01.
pdf
[38] ‘KAPLAN HYDRO GENERATOR UNIT’, GE Renewable Energy.
https://www.ge.com/renewableenergy/sites/default/files/2020-01/hydro-cgi-kaplan-
generating-unit-3000px.jpg (accessed May 15, 2021).
[39] ‘AC generator’, Synchronous machines Design of synchronous machines, Dec. 16, 2020.
https://www.freeenergyplanet.biz/renewable-energy-systems/synchronous-machines-
design-of-synchronous-machines.html (accessed Apr. 22, 2021).
[40] T. Øyvang, ‘Enhanced power capability of generator units for increased operational
security’, PhD, University of South-Eastern Norway (USN), Porsgrunn, 2018.
[41] E. Westgaard and O. Andersen, ‘Dimensjoneringseksempel for synkronmaskin’, NTH,
Department of Electrical machines, Trondheim Norway, p. 52, 1965.
[42] ‘IEEE Std 115TM-2009, IEEE Guide for Test Procedures for Synchronous Machines’,
IEEE, 2010.
[43] Chas. P. Steinmetz, ‘On the Law of Hysteresis’, Trans. Am. Inst. Electr. Eng., vol. IX, no.
1, pp. 1–64, 1892, doi: 10.1109/T-AIEE.1892.5570437.
[44] D. M. Ionel, M. Popescu, M. I. McGilp, T. J. E. Miller, S. J. Dellinger, and R. J.
Heideman, ‘Computation of Core Losses in Electrical Machines Using Improved Models
for Laminated Steel’, IEEE Trans. on Ind. Applicat., vol. 43, no. 6, pp. 1554–1564, 2007,
doi: 10.1109/TIA.2007.908159.

99
 References
[45] P. Parthasaradhy and S. V. Ranganayakulu, ‘Hysteresis and eddy current losses of
magnetic material by Epstein frame method-novel approach’, The International Journal
Of Engineering And Science (IJES), p. 9, 2014.
[46] Transformer Handbook, 3rd ed. ABB, 2007.
[47] ‘Rotating Electrical Machines Losses and Efficiency Classification’:, EEEGuide.
https://www.eeeguide.com/rotating-machines-losses-classification/
[48] J. Vranick, ‘Prediction of Windage Power Loss In alternators’, NASA Technical Note,
Cleveland, Ohio, United States, NASA-TN-D-4849, Oct. 1968. [Online]. Available:
https://ntrs.nasa.gov/search.jsp?R=19680027690
[49] D. Pejovski, B. Velkovski, and K. Najdenkoski, ‘MATLAB Model for Visualization of
PQ diagram of a Synchronous Generator’, ResearchGate, p. 5, Nov. 2016.
[50] P. Kundur, Power System Stability and Control. McGraw-Hill, Inc, 1993.
[51] E. Csanyi, ‘Large power transformer tailored to customers’ specifications’, Electrical
Engineering Portal (EEP), Dec. 30, 2013. https://electrical-engineering-portal.com/an-
overview-of-large-power-transformer-lpt (accessed May 15, 2021).
[52] E. Vaahedi, Practical Power System Operation. Hoboken, New Jersey: IEEE
Press/Wiley, 2014.
[53] E. D. C. Bortoni, ‘EFFICIENCY MAP AND WEIGHTED AVERAGE EFFICIENCY
FOR SYNCHRONOUS MACHINES FOR HYDROPOWER PLANTS’, FEDERAL
UNIVERSITY OF ITAJUBÁ, 2019.
[54] J. Arrillaga and C. P. Arnold, ‘Power System Stability - Advanced Component
Modelling’, in Computer Analysis of Power Systems, John Wiley & Sons, Ltd, 1990, p.
11.
[55] S. Beckwith, ‘Approximating Potier Reactance’, Electr. Eng., vol. 56, no. 7, pp. 813–818,
Jul. 1937, doi: 10.1109/EE.1937.6540240.
[56] I. Ilić, Maljković, and I. Gašparac, ‘Methodology For Determining The Actual PQ
Diagram Of A Hydrogenerator’, vol. 56, no. 2, pp. p144-181, 2007.
[57] E. S. Menon, Transmission Pipeline Calculations and Simulations Manual. Sint Louis:
Elsevier Science & Technology, 2015.
[58] H. Kudela, ‘Hydraulic losses in pipes’, Wroclas University of Science and Technology, p.
9, 2012.
[59] H. Belyadi, E. Fathi, and F. Belyadi, ‘Chapter Eight - Hydraulic Fracturing Chemical
Selection and Design’, in Hydraulic Fracturing in Unconventional Reservoirs, Elsevier
Inc., 2017, p. 13. [Online]. Available: https://doi.org/10.1016/B978-0-12-849871-
2.00008-3

100
 References

Appendix A
Task Description of the Master Thesis
This appendix contains the original task description and main goals for this thesis.

101
 References

Appendix B
Minutes of meeting, second formal project meeting
13.04.2021
This appendix contains the minutes of meeting from the second project meeting, where the task
description was formally changed.

102
References

103
References

104
 References

Appendix C
Theoretical study of losses in a waterway

Estimation of friction losses

Calculation methods for estimating friction losses in waterway have been developed by
numerous people, where each has its pros and cons. Hagen-Poiseuille is an analytical equation
used to calculate pressure drop through waterways, where the flow is assumed to be a
Newtonian fluid, laminar and incompressible. The assumptions in Hagen-Poiseuille will give
incorrect results where the flow in headrace tunnels is turbulent and have high frictional losses
due to the rough walls. Empirical formulas like Darcy-Weisbach’s and Manning’s equation are
developed to achieve suitable accuracy for large tunnels with uneven surfaces. The use of
Darcy-Weisbach’s and Manning’s equation to estimate head losses are recommended during
the planning and design of the system, where both provide similar accuracy for a given range
[25]. The Darcy-Weisbach would be used as an example, as this method is widely used [26].
Darcy-Weisbach’s equation expresses the losses in terms of head losses [m] as shown:
𝐿 ∙ 𝑣2
ℎ𝑓 = 𝑓 ∙ 10.1
2𝑔 ∙ 𝐷ℎ
where ℎ𝑓 = head loss [𝑚], 𝑓 = friction factor, 𝑣 = flow velocity [𝑚/𝑠], 𝐿 = tunnel length [𝑚],
𝑔 = gravitational constant [𝑚/𝑠 2 ] and 𝐷ℎ = hydraulic diameter [𝑚].
The friction factor 𝑓 represents friction losses in a pipe/tunnel section or a valve, dependent on
the type of flow; laminar or turbulent. Reynolds number is a measurement that determines the
magnitude of turbulence in fluids. In circular pipes, where the velocity and pipe diameter are
small (𝑅𝑒 < 2000), the flow is considered to be laminar or low turbulent [26], and the equation
for friction factor 𝑓 could be used as follows:
64
𝑓= 10.2
𝑅𝑒
Reynold number is defined by:
𝜌 ∙ 𝑣 ∙ 𝐷ℎ
𝑅𝑒 = 10.3
µ
where 𝑅𝑒 = Reynold number, 𝜌 = density of the fluid [𝑘𝑔/𝑚3 ], 𝑣 = velocity of the fluid [𝑚/𝑠],
𝐷ℎ = pipe diameter [𝑚] and µ = dynamic viscosity of the fluid [𝑘𝑔/(𝑚 ∙ 𝑠)].
When the Reynolds number is between 2000 and 4000, the flow is unstable, and turbulent and
laminar conditions could occur [26]. For turbulent flow, one can use the Colebrook-White, an
empirical formula used to estimate the friction factor used in the Darcy-Weisbach’s equation
based on Reynold’s number the relative surface roughness found tables. The Colebrook-White
equation, also referred to as the Colebrook equation, is defined as [57] [58]:
1 𝜀 2.51
= −2 log10 ∙ ( + ) 10.4
√𝑓 3.7𝐷ℎ 𝑅𝑒 √𝑓

105
 References
where, 𝑓 = friction factor solved numerically, 𝜀 = absolute surface roughness [𝑚𝑚] and 𝐷ℎ =
pipe diameter [𝑚].
It is essential to distinguish between absolute and relative roughness. Relative roughness 𝑒
expresses the amount of roughness existing inside a pipe and is defined as the absolute
roughness 𝜀 divided by the inside diameter 𝐷 of a pipe, yielding [59]:
𝜀
𝑒= 10.5
𝐷

A Moody diagram shown in the figure below, is an alternative method to estimate the friction
factor when the relative roughness and Reynold’s number is known.

Moody diagram, a diagram used to determine friction factor (f) from the relative roughness
factor and the Reynolds number [58].

106
 References
Table for determining local losses in waterway
Typical resistance coefficients (k) for pipe fittings [26].

107
 References

Appendix D
Derivation of head loss formula with a total head loss
coefficient
An equation for head loss can be derived from the equations representing friction losses (3.1)
and local losses (3.2). The following method has been used:

The relationship between volume discharge Q and velocity is:

𝑄𝑓𝑙𝑜𝑤 4
𝑣= = 𝑄𝑓𝑙𝑜𝑤 10.6
𝐴 𝜋𝐷ℎ2

Substituting (10.6) into the Darcy-Weisbach’s equation (10.1) and introducing the head loss
coefficient 𝑘𝑓,𝑖 , which yields:
2
ℎ𝑙𝑜𝑠𝑠,𝑖 = 𝑘𝑓,𝑖 𝑄𝑓𝑙𝑜𝑤 10.7

where 𝑘𝑓,𝑖 is the head loss coefficient for friction loss to each component or section of the
waterway and is given by:
2
𝐿𝑖 4 𝐿𝑖
𝑘𝑓,𝑖 =𝑓∙ ∙( ) = 𝑓𝑖 ∙ 3
10.8
2𝑔 ∙ 𝐷ℎ,𝑖 𝜋𝐷ℎ,𝑖 8𝑔𝜋 2 𝐷ℎ,𝑖

In a similar manner as the head loss coefficient based on friction losses (10.7 and 10.8), the
head loss coefficient for local losses will be given by substituting (10.6) into (3.2), which
yields:
2
ℎ𝑙𝑜𝑠𝑠,𝑗 = 𝑘𝑙,𝑖 𝑄𝑓𝑙𝑜𝑤 10.9

and
2
kj 4
𝑘𝑙,𝑖 = ( 2 ) 10.10
2g 𝜋𝐷ℎ,𝑗

where 𝑘𝑙,𝑖 is the head loss coefficient for local losses, 𝑘𝑗 is the resistance coefficient and 𝐷ℎ,𝑗
is the part diameter.
Summation of all the head loss coefficients 𝑘𝑓,𝑖 and 𝑘𝑙,𝑖 for 𝑖 = 1,2 … , 𝑛 and 𝑗 = 1,2 … , 𝑚,
respectfully, would produce a single coefficient K representing the total head loss, expressed
as:
𝑛 𝑚
2 2
𝐻𝑙𝑜𝑠𝑠 = 𝑄𝑓𝑙𝑜𝑤 (∑ 𝑘𝑓,𝑖 + ∑ 𝑘𝑙,𝑗 ) = 𝐾𝑄𝑓𝑙𝑜𝑤 10.11
𝑖=1 𝑗=1

Illustration of head losses shown in sections.

108
 References

Appendix E
Calculation example of field current
Input parameters to the generator: 𝑆𝑛 = 103 MVA, 𝑉𝑛 = 11 𝑘𝑉, 𝑋𝑑 = 1.06 Pu, 𝑋𝑞 = 0.69
Pu, 𝑋𝑑′ = 0.15 Pu, 𝑋𝑙 = 0.11 Pu, 𝑅𝑎 = 0.0027 Pu
Operation data: 𝑃 = 0.900 Pu, 𝑄 = 0.436 Pu, 𝑉 = 1 Pu
Field current is calculated with the following method, based on [53] and [54]:
√𝑃𝑡2 +𝑄𝑡2 √0.900 2 +0.4362
1. 𝐼𝑡 = = = 1 𝑃𝑢
𝑉𝑡 1
𝑄 0.436
2. 𝜑 = tan−1 ( 𝑃𝑡) = tan−1 (0.900) = 0.45 (𝑟𝑎𝑑) 𝑜𝑟 0.9 (𝑑𝑒𝑔)
𝑡

𝑋𝑞 𝐼𝑡 𝑐𝑜𝑠𝜑−𝑅𝑎 𝐼𝑡 𝑠𝑖𝑛𝜑 0.69∙1∙cos(0.45)−0.0027∙1∙sin (0.45)

3. 𝛿𝑖 = tan−1 (𝑉 +𝑅 ) = tan−1 ( ) = 0.44
𝑡 𝑎 𝐼𝑡 𝑐𝑜𝑠𝜑+𝑋𝑞 𝐼𝑡 𝑠𝑖𝑛𝜑 1+0.0027∙1∙cos(0.45)+0.69∙1∙sin (0.45)

4. 𝐸𝑔 = 𝑉𝑡 𝑐𝑜𝑠𝛿𝑖 + 𝑅𝑎 𝐼𝑡 𝑐𝑜𝑠(𝜑 + 𝛿𝑖 ) + 𝑥𝑑 𝐼𝑡 𝑠𝑖𝑛(𝜑 + 𝛿𝑖 )

𝐸𝑔 = 1 ∙ cos(0.44) + 0.0027 ∙ 1 ∙ 𝑐𝑜𝑠(0.45 + 0.44) + 1.06 ∙ 1 ∙ 𝑠𝑖𝑛(0.45 + 0.44)
𝐸𝑔 = 1.73
5. Air-gap line: 𝐸𝑔,𝑜 is the excitation voltage at open circuit.
𝐸𝑔,𝑜 1
𝑏𝑣 = = 556 = 0.0018
𝐼𝐹

Open circuit characteristic (OCC) drawn in blue and air-gap line drawn in purple.
𝐸𝑔 1.73
6. 𝐼𝐹𝑈 = = 0.0018 = 961.1 [𝐴]
𝐵𝑣

7. 𝑋𝑝 = 𝑋𝑙 + 0.63(𝑋𝑑′ − 𝑋𝑙 ) = 0.11 + 0.63(0.15 − 0.11) = 0.135 Pu

109
 References
𝑋𝑝 𝐼𝑡 𝑐𝑜𝑠𝜑−𝑅𝑎 𝐼𝑡 𝑠𝑖𝑛𝜑 0.135 ∙1∙cos(0.45)−0.0027∙1∙sin (0.45)
8. 𝛿𝑝 = tan−1 (𝑉 +𝑅 ) = tan−1 ( )
𝑡 𝑎 𝐼𝑡 𝑐𝑜𝑠𝜑+𝑋𝑝 𝐼𝑡 𝑠𝑖𝑛𝜑 1+0.0027∙1∙cos(0.45)+0.135 ∙1∙sin (0.45)

𝛿𝑝 = 0.113
9. 𝐸𝑝 = 𝑉𝑡 𝑐𝑜𝑠𝛿𝑝 + 𝑅𝑎 𝐼𝑡 𝑐𝑜𝑠(𝜑 + 𝛿𝑝 ) + 𝑥𝑝 𝐼𝑡 𝑠𝑖𝑛(𝜑 + 𝛿𝑝 )
𝐸𝑝 = 1 ∙ cos(0.113 ) + 0.0027 ∙ 1 ∙ 𝑐𝑜𝑠(0.45 + 0.113 ) + 1.06 ∙ 1 ∙ 𝑠𝑖𝑛(0.45 +
0.113 )
𝐸𝑝 = 1.068
10. 𝐼𝐹𝑃 = (∑𝑛𝑖=1 𝑏𝑖 𝐸𝑝𝑖 ) = (𝐸𝑝 + 𝐶𝑛 𝐸𝑝𝑛 )𝑘

Dotted blue curve represent the actual OCC whereas the blue curve represents the fitted curve. Purple line
represents the air-gap line
Find the coefficient 𝐶𝑛 , 𝑛 and 𝑘:
𝑘 = 𝐼𝐹,𝑛 = 556
𝑛 is usually set to 7 or 9.
𝐶𝑛 is found by plotting the relationship between the calculated field current (𝐼𝐹𝑃 ) and
the voltage (e.g., 0-1.25) several times for different values of 𝐶𝑛 until the curve is close
to the measured OCC, as depicted above.
For this scenario: 𝑘 = 556, 𝑛 = 9, 𝐶𝑛 = 0.11
𝐼𝐹𝑃 = (1.068 + 0.11 ∙ 1.0689 ) ∙ 556 = 704.4 [𝐴]
𝐸𝑝 1.068
11. 𝐼𝐹𝑆 = 𝐼𝐹𝑃 − 𝑏 = 704.4 − 0.0018 = 111.1 [𝐴]
𝑣

12. 𝐼𝑓𝑑 = 𝐼𝐹𝑈 + 𝐼𝐹𝑆 = 961.1 [𝐴] + 111.1 [𝐴] = 1072.2 [𝐴]

110
 References

Appendix F
Hydropower specifications and parameters for
Åbjøra

111
References

112
References

113
References

114
 References
Turbine data from Åbjøra:

Gross head (nominal) = 445.3 m

Turbine Water
power discharge Efficiency
MW 3 %
𝑚 /𝑠
38.35 10.13 88.9
48.14 12.39 91.23
63.44 15.94 93.45
75.51 18.8 94.35
81.21 20.18 94.53
85.28 21.18 94.56
91.98 22.87 94.46
98.13 24.47 94.18

Gross head = 442.6 m

Turbine Water
power discharge Efficiency
MW 3 %
𝑚 /𝑠
38.81 10.18 88.95
48.7 12.46 91.26
64.17 16.03 93.46
76.36 18.89 94.35
82.12 20.28 94.53
86.24 21.29 94.56
93 22.98 94.45
99.21 24.59 94.17

Gross head = 439.2 m

Turbine Water
power discharge Efficiency
MW 3 %
𝑚 /𝑠
37.75 10.06 88.83
47.4 12.31 91.18
62.49 15.84 93.42
74.4 18.67 94.34
80.03 20.05 94.53
84.05 21.05 94.56
90.65 22.72 94.47
96.72 24.31 94.19

115
 References

Input parameters
Parameter Value

Waterway

Nominal gross head 442 m

𝑘𝑎 (Total head loss coefficient K, 0.016335

waterway)

Generator

𝑆𝑛 (Nominal apparent power, generator) 103 MVA

𝑉𝑛 (Nominal voltage) 11 kV

𝑅𝑓 (Field winding resistance, 75℃) 0.15253 Ω

𝑅𝑎 (Armature winding resistance, 75℃) 0.003155 Ω

𝑋𝑑 (Synchronous reactance d-axis) 1.06 Pu

𝑋𝑞 (Synchronous reactance q-axis) 0.69 Pu

𝑘𝑒𝑥 (excitation constant) 1.116

𝑋𝑑′ (Transient reactance) 0.15 Pu

𝑋𝑙 (Leakage reactance) 0.11 Pu

𝑏_𝑣 (Air-gap reactance) 0.0018

𝑛 (generator OCC exponent) 9

𝐶𝑛 (Polynomial constant, OCC) 0.12

𝐼𝑓𝑑,𝑛 (Nominal field current) 556 A

𝑃𝐹𝐸 (Core losses) 0.21192 MW

𝑃𝑉 (windage and ventilation) 0.17292 MW

𝑃𝐹 (Bearing losses, generator) 0.24090 MW

Transformer

Nominal transformer rating 103 MVA

𝑃0 (No-load losses) 0.0525 MW

𝑃𝑘 (Load losses) 0.225 MW

116
 References

Appendix G
Hydropower specifications and parameters for
Sundsbarm

Characteristics of the generator in Sundsbarm: Open circuit characteristic (OCC), air-gap line
and estimated field current relative to armature current (nominal terminal voltage).

117
 References
Generator data from Sundsbarm (nominal)
𝒄𝒐𝒔𝝋 =0.85

Load % 100

Power KVA 118 000

Stator current A 4542.0

Mechanical losses kW 537

Core losses kW 353

Load losses kW 333

Total excitation losses kW 225

Total losses kW 1450

P out kW 100 300

P in kW 101 750

Efficiency % 98.575

Calculation of excitation losses

Field current A 1001
2
𝐼𝑓𝑑 ∙ 𝑅(75℃) Loss kW 205.45
Excitation and brush losses kW 19.55
Total excitation losses kW 225
R field at 20℃ 75℃
Field resistance 0.20504 Ω 0.24857 Ω
Armature resistance 0.00322Ω 0.00390 Ω

118
 References
Turbine data from Sundsbarm:
Gross head = 465.5 m
Turbine power Water discharge Efficiency
MW 3 %
𝑚 /𝑠
33.993 8.7741 85.832
47.048 11.643 89.527
55.854 13.575 91.156
68.021 16.259 92.688
76.869 18.236 93.389
86.809 20.5 93.813
97.158 22.94 93.83
105.24 24.92 93.563

Gross head = 473.7 m

Turbine power Water discharge Efficiency
MW 3 %
𝑚 /𝑠
35.177 8.9043 86.024
48.65 11.815 89.664
57.735 13.775 91.267
70.284 16.497 92.77
79.408 18.502 93.454
89.655 20.799 93.862
100.32 23.273 93.865
108.64 25.277 93.593

Gross head (nominal) = 493 m

Turbine power Water discharge Efficiency
MW 3 %
𝑚 /𝑠
38.005 9.2073 86.366
52.475 12.215 89.885
62.226 14.24 91.43
75.686 17.052 92.87
85.465 19.122 93.516
96.44 21.493 93.886
107.87 24.046 93.862
116.79 26.114 93.58

119
 References

Input parameters
Parameter Value

Waterway

Nominal gross head 490 m

𝑘𝑎 (Total head loss coefficient K, 0.00893

waterway)

Generator

𝑆𝑛 (Nominal apparent power, generator) 118 MVA

𝑉𝑛 (Nominal voltage) 15 kV

𝑅𝑓 (Field winding resistance, 75℃) 0.24857 Ω

𝑅𝑎 (Armature winding resistance, 75℃) 0.00396 Ω

𝑋𝑑 (Synchronous reactance d-axis) 1.13 Pu

𝑋𝑞 (Synchronous reactance q-axis) 0.61 Pu

𝑘𝑒𝑥 (excitation constant) 1.1

𝑋𝑑′ (Transient reactance) 0.29 Pu

𝑋𝑙 (Leakage reactance) 0.12 Pu

𝑏_𝑣 (Air-gap reactance) 0.00196

𝑛 (generator OCC exponent) 9

𝐶𝑛 (Polynomial constant, OCC) 0.075

𝐼𝑓𝑑,𝑛 (Nominal field current) 509 A

𝑃𝐹𝐸 (Core losses) 0.353 MW

𝑃𝑉 (windage, ventilation and bearing) 0.537 MW

Transformer

Nominal transformer rating 118 MVA

𝑃0 (No-load losses) 0.1484 MW

𝑃𝑘 (Load losses) 0.2814 MW

120
 References

Appendix H
DATA FORMATTING:
The procedure for formatting the data is performed in excel and described in four steps where
the original data set will be gradually improved until the finished format in step four.
- Step 1: Convert the data set from “NaN” values to numbers.
- Step 2: Add data (assumed) points at locations with no numbers (only the first few
days).
- Step 3: Convert data from 1-hour resolution to 15-minute semi-resolution to improve
the accuracy in start/stop regions.
- Step 4: Estimate the data in the 15-minute semi resolution.

1. In the data set, there are originally numerous cells with “NaN”, which says that there is
no change in value relative to the previous cell. This “NaN” value must be replaced
with actual values in order to do the calculation. The procedure to convert the data set
is as follows:
I. Converts all “NaN” values with a unique constant like “999999”
II. Use the excel command for all cells for 𝐴𝑜𝑙𝑑 ∈ (1,n):
=If(𝐴𝑜𝑙𝑑 <999999; 𝐴𝑜𝑙𝑑 ; 𝐴𝑛𝑒𝑤,𝑖−1 )
where 𝐴𝑜𝑙𝑑 is a value (P,Q or V) at row (1 - n), and 𝐴𝑛𝑒𝑤,𝑖−1 is the previously
calculated value (0 – n-1) illustrated in the table below. The method requires an
initial value, namely 𝐴𝑛𝑒𝑤,0 = 𝐴𝑜𝑙𝑑,0 .

𝐴𝑜𝑙𝑑,0 𝐴𝑛𝑒𝑤,0

𝐴𝑜𝑙𝑑,1 𝐴𝑛𝑒𝑤,1

𝐴𝑜𝑙𝑑,2 𝐴𝑛𝑒𝑤,2

𝐴𝑜𝑙𝑑,3 𝐴𝑛𝑒𝑤,3

Exception for Åbjøra HPU: NaN is used differently where:

▪ Active power (P) and reactive power (Q): NaN = 0
▪ Voltage (V): NaN = 0 or 11 kV.
With these findings, one could determine if the NaN at the voltage should equal
0 or 11 by comparing the with the active power (P). The excel command will
therefore be:
𝑉𝑛𝑒𝑤,𝑖 = 𝐼𝐹(𝐴𝑁𝐷(𝑉𝑖 = 9999; 𝑃𝑖 > 0); 11; 𝐼𝐹(𝑉𝑖 < 11; 𝑉𝑖 ; 0))

121
 References
where 𝑉𝑛𝑒𝑤,𝑖 is a new column for the filtered voltage column, 𝑉𝑖 is the original
voltage column, where NaN is first replaced with a random value like 9999 in
order for performic the numeric command, 𝑃𝑖 is the column for active power (P)
and 11 is representing the nominal voltage 11 kV.
2. Corrected for logical errors where it seems like the generator have multiple start/stop
sequences between two stable regions, and these regions will be set to zero. A stable
region is regarded as a region where the voltage is above a trigger point or a voltage
that is considered close to “normal”.
I. The table will be corrected for logical errors where it seems like the generator
have multiple start/stop operations during few hours, as shown below:
𝑖 Voltage [kV] 𝑉𝑖,𝑜𝑙𝑑 Voltage [kV] 𝑉𝑖,𝑛𝑒𝑤
1 14.3 14.3
2 14.4 14.4
3 6.4 6.4
4 6.4 0
5 6.4 6.4
6 14.3 14.3
The generator will be regarded as disconnected (off) when the voltages close to
the measurement are substantially lower than the rated voltage. The excel
command for this correction is as follows:
𝑉𝑖,𝑛𝑒𝑤 = 𝐼𝐹(𝐴𝑁𝐷(𝑉𝑖−1,𝑜𝑙𝑑 < 𝑉𝑡𝑟𝑖𝑔 ; 𝑉𝑖,𝑜𝑙𝑑 < 𝑉𝑡𝑟𝑖𝑔 ; 𝑉𝑖+1,𝑜𝑙𝑑 < 𝑉𝑡𝑟𝑖𝑔 ; ); 0; 𝑉𝑖,𝑜𝑙𝑑
where 𝑉𝑥 is the cell number (voltage) and 𝑉𝑡𝑟𝑖𝑔 is the minimum voltage
recognised as a possible non-operation situation. For the active power (P) and
reactive power (Q), the command is quite simple:
𝑋𝑖 = 𝐼𝐹($𝐴𝑖 = 0; 0; 𝑋𝑖 )
which states that P or Q (𝑋𝑖 ) shall be 0 if the voltage (𝑉𝑖 ) is zero, else P or Q
shall remain unchanged.

3. Since each value in the data set is an average value measured over one hour, there are
unrealistically low measurements. An example could be a voltage measurement of 5
kV (average) occurring at a generator with a nominal voltage of 10 kV. The actual
situation would probably be 10 kV for the first 30 minutes and then 0 kV for the next
30 minutes, giving an average of 5 kV.
In order to have a compatible data set with the MATLAB program, each time step must
have the same length. Thereby, it will be created a new column where each data point
is used four times, i.e., time steps of 15 minutes.
I. A new row will be created. This will be used for counting each value four times.
Cell nr. C
1 1
2 1
… …
4 1

122
 References
5 = 𝐶1 +1
… …
𝑛 = 𝐶𝑛−4+1

II. With the help of the counting row previously made, and the following excel
command, copy the value in column “A” and paste the same value in four rows
in the “B” column, shown in the table below. (A and B are just cells representing
old values of P,Q and V, se table below).
𝐵𝑗 = 𝐼𝑁𝐷𝐼𝑅𝐸𝐶𝑇("A" & $𝐶𝑖 )
The “INDIRECT” function will create a value in 𝐵𝑗 equal to 𝐴𝑖 , where “A” is
nothing more than the column name in excel and 𝐶𝑖 the number in counting row.
The first 𝐶𝑖 will be 1,1,1,1,2,2,2,2,3…n, as shown in the table below. The
number of data rows in data set (1 hour) created in (3,Ι), have a length of 𝑛 =
8746, and the new data set (15 minutes resolution) will have a length 𝑚 = 4 ∙
𝑛.
Data set Data set Counting
(1 hour) (15 minutes) row
A B C
𝐴𝑜𝑙𝑑,1 𝐴𝑛𝑒𝑤,1 1
𝐴𝑜𝑙𝑑,2 … …
𝐴𝑜𝑙𝑑,3 𝐴𝑛𝑒𝑤,1 1
𝐴𝑜𝑙𝑑,4 𝐴𝑛𝑒𝑤,2 2
𝐴𝑜𝑙𝑑,5 … …
𝐴𝑜𝑙𝑑,6 𝐴𝑛𝑒𝑤,2 2
… … …
𝐴𝑜𝑙𝑑,𝑛 𝐴𝑛𝑒𝑤,𝑚 𝑛

4. From the table with 15 minutes resolution, there will be added two additional columns.
These columns will be used to determine how many of the four (15 minutes) operations
that will be used.
I. The first column represents the probable duration of an operation (0-1) given by
the voltage ratio:
𝑉𝑖
𝑑𝑢𝑟𝑎𝑡𝑖𝑜𝑛 =
𝑉𝑛𝑜𝑟𝑚𝑎𝑙
where 𝑉𝑖 is the operating voltage and 𝑉𝑛𝑜𝑟𝑚𝑎𝑙 is the voltage with a high
probability of occurrence (14.6 kV for Sundsbarm).

II. The second column is only a repetitive pattern named “REP” and is as follows
[0.25 0.50 0.75 1.00 0.25 ...]. The column is used in combination with the
duration (4, Ι) to determine the number of operations (1-4) to be used.

123
 References
III. The final step is to apply the “Duration” and “Rep” columns into the table from
(3, ΙΙ).
The full table will look like:
V P Q Duration Rep 𝑉𝑓𝑖𝑛𝑎𝑙 𝑃𝑓𝑖𝑛𝑎𝑙 𝑄𝑓𝑖𝑛𝑎𝑙
[kV] [MW] [MVAr] [%]
𝐴1 𝐴2 𝐴3 𝐷 𝐸 AA BB CC
14.6 102.0 18.1 1 0.25 14.6 102.0 18.1
14.6 102.0 18.1 1 0.50 14.6 102.0 18.1
14.6 102.0 18.1 1 0.75 14.6 102.0 18.1
14.6 102.0 18.1 1 1.00 14.6 102.0 18.1
8.0 101.5 15.0 0.5 0.25 14.6 101.5 15.0
8.0 101.5 15.0 0.5 0.50 14.6 101.5 15.0
8.0 101.5 15.0 0.5 0.25 0 0 0
8.0 101.5 15.0 0.5 0.50 0 0 0
From the table above, the final table will be created with this excel command:

𝑋𝑣𝑎𝑟,𝑖 = 𝐼𝐹 (𝐷𝑖 > 0.75; 𝑋𝑖 ; 𝐼𝐹 (𝑂𝑅 (𝐴𝑁𝐷(𝐷𝑖 = 0.5; 𝐸𝑖 < 0.75; 𝐴𝑁𝐷(𝐷𝑖 <

0.75; 𝐷𝑖 > 0.5; 𝐸𝑖 < 1)); 𝑋𝑖 ; 𝐼𝐹(𝐴𝑁𝐷(𝐷𝑖 > 0.25; 𝐸𝑖 < 0.5; 𝐸𝑖 <

0.75); 𝑋𝑖 ; 𝐼𝐹(𝑂𝑅(𝐷𝑖 < 0.25; 𝐷𝑖 = 0.25); 0; 0))))

𝐴𝐴𝑖 = 𝐼𝐹 (𝐴𝑁𝐷(𝑋𝑣𝑎𝑟,𝑖 < 𝑉𝑡𝑟𝑖𝑔 ; 𝑋𝑣𝑎𝑟,𝑖 > 0); 𝑉𝑡𝑟𝑖𝑔 ; 𝐼𝐹 (𝑋𝑣𝑎𝑟,𝑖 =

0; 0; 𝐼𝐹(𝑂𝑅(𝑋𝑣𝑎𝑟,𝑖 = 𝑉𝑡𝑟𝑖𝑔 ; 𝑋𝑣𝑎𝑟,𝑖 > 𝑉𝑡𝑟𝑖𝑔 ); 𝑋𝑣𝑎𝑟,𝑖 ; 0)))

𝐵𝐵𝑖 = 𝐼𝐹 (𝐷𝑖 = 1; 𝑋𝑣𝑎𝑟,𝑖 ; 𝐼𝐹 (𝐴𝑁𝐷(𝐷𝑖 < 1; 𝑋𝑣𝑎𝑟,𝑖 = 𝑃𝑡𝑟𝑖𝑔,𝑚𝑎𝑥 ; 𝑋𝑣𝑎𝑟,𝑖 >

𝑃𝑡𝑟𝑖𝑔,𝑚𝑎𝑥 ); 𝑋𝑣𝑎𝑟,𝑖 ; 𝐼𝐹 (𝐴𝑁𝐷(𝐷𝑖 < 1; 𝑋𝑣𝑎𝑟,𝑖 > 𝑃𝑡𝑟𝑖𝑔,𝑚𝑖𝑛 ; 𝑋𝑣𝑎𝑟,𝑖 <

𝑃𝑡𝑟𝑖𝑔,𝑚𝑎𝑥 ); 𝑃𝑡𝑟𝑖𝑔,𝑚𝑎𝑥 ; 𝐼𝐹(𝑂𝑅(𝑋𝑣𝑎𝑟,𝑖 < 𝑃𝑡𝑟𝑖𝑔,𝑚𝑖𝑛 ; 𝑋𝑣𝑎𝑟,𝑖 = 𝑃𝑡𝑟𝑖𝑔,𝑚𝑖𝑛 ); 0; 0))))

𝐶𝐶𝑖 = 𝑋𝑣𝑎𝑟,𝑖
where 𝑋𝑣𝑎𝑟,𝑖 is a variable to determine the value based on the duration, used
here to make the command more readable (and used directly for reactive power),
𝑋𝑖 ∈ [𝐴𝑖 ,𝐵𝑖 ,𝐶𝑖 ], 𝑉𝑡𝑟𝑖𝑔 is the minimum voltage recognised as a possible non-
operation situation (used in (3, Ι), 𝐴𝐴𝑖 , 𝐵𝐵𝑖 , 𝐶𝐶𝑖 are column vectors for voltage
(V), active power (P) and reactive power (Q), respectfully, 𝑃𝑡𝑟𝑖𝑔,𝑚𝑎𝑥 is the
assumed active power for all operations where 𝐷𝑖 < 1 and P is in the range
between 𝑃𝑡𝑟𝑖𝑔,𝑚𝑎𝑥 and 𝑃𝑡𝑟𝑖𝑔,𝑚𝑖𝑛 , and 𝑃𝑡𝑟𝑖𝑔,𝑚𝑖𝑛 is the absolute minimum active
power any operation can be. In addition, the reactive power will may result in

124
 References
values different from zero when both P and V are zero. This was solved by
setting Q = 0 if P = 0.
Explained in short:
o If the voltage is above trigger voltage, then the V, P and Q will be on for 4
periods (60 min), and all values remains the same.
o If the voltage is under trigger voltage, then the duration will determine the
period for which V, P and Q are on. The voltage will also be set to the trigger
voltage for the “on” period.
▪ If active power P is below min trigger P, then P = 0.
▪ If active power P is above min trigger P and below Max trigger P, then
P = max trigger P.
▪ If active power P is above max trigger P, then P = P (old)
o Q will always be set to the original value, but will be zero for all periods where
P is zero.
To summarise, the assumed values for Åbjøra and Sundsbarm are:
Normal Trigger Max trigger P Min trigger P
voltage voltage
Åbjøra 11 10 45 20
Sundsbarm 14.6 14.4 75 20

125
 References

Appendix I
Additional results from Åbjøra

126
References

127
 References

Appendix J
Additional results from Sundsbarm

Efficiency mapping of the entire HPU

128
 References

Efficiency mapping of the generator in Sundsbarm

Efficiency mapping of the Francis turbine in Sundsbarm

129
 References

Appendix K
Function description for the MATLAB program.

PQ_diagram function:
Description: This function pots the PQ-diagram with:
- Armature current limit
- Field current limit
- Generator maximum power (active power limit)
- Practical stability limit
- Minimum field current limit
Assumptions:
- Practical stability is assumed to start where operational, and armature
field current intersects and ends where practical stability limit crosses
the minimum field current limit.
- The minimum field current is calculated with a minimum excitation
voltage assumed to be 𝐸𝑚𝑖𝑛 = 𝑘𝑚𝑖𝑛 ∙ 𝐸𝑚𝑎𝑥 , where 𝑘𝑚𝑖𝑛,𝑛𝑜𝑚 = 0.3.
(𝑘𝑚𝑖𝑛 is usually between 0-0.3)
The variable fig_n is only used to combine the PQ diagram to other figures.
Input parameters
PF, Xq, Xd, fig_n

130
 References
PQ_diagram function:
Description: This function pots the PQ-diagram with:
- Armature current limit
- Field current limit
- Turbine maximum operational limit
- Practical stability limit
- Minimum field current limit
Assumptions:
- Turbine maximum operational limit = nominal active power of the
generator.
- Practical stability is assumed to start where operational, and armature
field current intersects and ends where practical stability limit crosses
the minimum field current limit.
- The minimum field current is calculated with a minimum excitation
voltage assumed to be Emin = k min ∙ Emax , where k min,nom = 0.3.
(k min is usually between 0-0.3)
Input parameters
PF, Xq, Xd

131
 References
Data function:
Description: Collect data from an excel file, and all input variables must have the same length.
There is an additional feature in the function that remove all data points where the
active power P is zero. This is done to eliminate zero division and would not
influence the program in any practical manner.

Input variables
Active power P [MW] Reactive power Q[MVAr] Voltage [kV] System Price [NOK/MWh]
Output variables
Active power P [MW] Reactive power Q[MVAr] Voltage [kV] System Price [NOK/MWh]

HPU function:
Description: The function calulates power losses and efficienies in the HPU, which are
determined by the input variables. Input variables could be given as a single value
or as a vector. The HPU function uses the “Field current” function to estimate the
field current. In addtion, function named “Turbine” which is a 2D interpolation
funciton.
Input parameters
Generator: Sn, Vn, Xq, Xd, Ra, Rf, k_ex, P_FE, P_V, P_F, bv, Xd_tran, X_leak, n_OCC, C_OCC, Ifd_n
Waterway: k_a
Turbine: Tur_P0, Tur_N0
Transformer: P0_Tran, Pk_Tran, Sn_Tran
Input variables
Generator: P [MW], Q [MVAr], V [kV]
Waterway: Head_gross [m]
Output variables (Efficiency (N) and Power loss (Ploss [MW])
Total HPU: Ploss_tot, Total_N
Transformer: Tran_N, Ploss_Tran
Generator: Gen_N, Ploss_Gen
Turbine: Tur_N, Ploss_Tur
Waterway: Water_N, Ploss_Water

Field current function:

Description: Determines the field current in the generator based on OCC.
Input parameters
Sn, Vn, Ra, Xq, Xd, bv, Xd_tran, X_leak, n_OCC, C_OCC, Ifd_n
Input variables
P [MW], Q [MVAr], V [kV]
Output variables
Ifd [kA]

132
 References

PQ_diagram function:
Description: This function pots the PQ-diagram with:
- Armature current limit
- Field current limit
- Generator maximum power (active power limit)
- Practical stability limit
- Minimum field current limit
Assumptions:
- Practical stability is assumed to start where operational, and armature
field current intersects and ends where practical stability limit crosses
the minimum field current limit.
- The minimum field current is calculated with a minimum excitation
voltage assumed to be 𝐸𝑚𝑖𝑛 = 𝑘𝑚𝑖𝑛 ∙ 𝐸𝑚𝑎𝑥 , where 𝑘𝑚𝑖𝑛,𝑛𝑜𝑚 = 0.3.
(𝑘𝑚𝑖𝑛 is usually between 0-0.3)
The variable fig_n is only used to combine the PQ diagram to other figures.
Input parameters
PF, Xq, Xd, fig_n

133
 References
Symbol MATLAB Description

Waterway

𝐾 k_a The total head loss coefficient

𝐻𝑔𝑟𝑜𝑠𝑠 Head_gross Gross head (total water head) [m]

Turbine

𝑃𝑡𝑢𝑟𝑏𝑖𝑛𝑒,𝑖 Tur_P0 A vector representing the output power of the turbine [MW]

𝜂𝑡𝑢𝑟𝑏𝑖𝑛𝑒,𝑖 Tur_N0 A vector representing the efficiency to the turbine at a given power [%]

Generator

𝑆𝑛 Sn Nominal apparent power [MVA]

𝑉𝑛 Vn Nominal voltage [kV]

𝑐𝑜𝑠𝜑 PF Nominal power factor

𝑋𝑞 Xq Synchronous reactance q to the machine [Pu]

𝑋𝑑 Xd Synchronous reactance d to the machine [Pu]

𝑅𝑎 Ra Armature winding resistance per phase [Ω]

𝑅𝑓𝑑 Rf Field winding resistance [Ω]

𝑋𝑎𝑑 Xad Equivalent inductance between excitation voltage and field current [Pu]

𝑘𝑒𝑥 k_ex Additional excitation loss factor, proportional to field winding losses

𝑃𝐹𝐸 P_FE Magnetization (iron) loss [MW]

𝑃𝑉 P_V Ventilation and windage loss [MW]

𝑃𝐹 P_F Friction loss in bearing and brushes [MW]

Transformer

𝑃0 P0_Tran Magnetization (iron) loss [MW]

𝑃𝑘 Pk_Tran Short circuit loss [MW]

𝑆𝑛,𝑡𝑟𝑎𝑛 Sn_Tran Nominal apparent power [MVA]

134
 References

Appendix L
MATLAB codes:
Hydropower (HPU) model:
function
[Ploss_tot,Total_N,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen,Tur_N,Ploss_Tur,Water_N,Ploss_Water,Q_flo
w_h] = HPU(P,Q,V,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv, Xd_tran, X_leak, n_OCC, C_OCC, Ifd_n,
Head_gross,k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran)
% Transformer Model:
P_Tran = P; %Input power
Ploss_Tran = P0_Tran + Pk_Tran.*(P_Tran./Sn_Tran).^2;
Tran_N = P./(P + Ploss_Tran);

% Generator Model:
P_G = P./Tran_N;
Q_G = Q;
% Load dependent losses (ohmic losses)
Ia = sqrt(P_G.^2+Q.^2)./(sqrt(3).*V); %[kA]
Ifd = Field_current(P, Q, V, Sn, Vn, Ra, Xq, Xd, bv, Xd_tran, X_leak, n_OCC, C_OCC,
Ifd_n); %[kA]
P_S = 3*Ia.^2*Ra; % Stator loss %[MW]
P_R = real(k_ex*Ifd.^2*Rf); % Rotor loss %[MW]
Ploss_Gen = P_S + P_R + P_FE + P_V + P_F;
Gen_N = P_G./(P_G+Ploss_Gen);

% Turbine Model:
P_Tur = P./(Tran_N.*Gen_N);
if Head0 == 0 %Simplified method (1D interpolation)
Tur_N = interp1(Tur_P0,Tur_N0,P_Tur, 'spline')/100;
Ploss_Tur = P.*(1-Tur_N);
else %Advanced method (2D interpolation)
% Sort the given heads, load and efficeincy:
len_P00 = length(Tur_P0(1,:));
% Place all data points of power in one vector
P0_Tur(1,1:len_P00) = Tur_P0(1,:);
P0_Tur(1,len_P00+1:2*len_P00) = Tur_P0(2,:);
P0_Tur(1,2*len_P00+1:3*len_P00) = Tur_P0(3,:);
P0_Tur = P0_Tur.';
% Create a matrix (3 X length_P00) of all heads
H0_Tur = ([linspace(Head0(1),Head0(1),len_P00),
linspace(Head0(2),Head0(2),len_P00), linspace(Head0(3),Head0(3),len_P00)]).';

N0_Tur(1,1:len_P00) = Tur_N0(1,:);
% Place all data points of efficiency in one vector
N0_Tur(1,len_P00+1:2*len_P00) = Tur_N0(2,:);
N0_Tur(1,2*len_P00+1:3*len_P00) = Tur_N0(3,:);
N0_Tur = N0_Tur.';
% Collect Turbine efficiency "Tur_N". Hm and Pm represents head and
% loading matrix (meshgrid) which can be used to plot turbine efficiency
[Tur_N, Hm, Pm] = Turbine(Head_gross, P_Tur, P0_Tur,H0_Tur,N0_Tur);
Ploss_Tur = P.*(1-Tur_N);
end

% Waterway Model:
P_W = P./(Tran_N.*Gen_N.*Tur_N);
Water_N = zeros(length(Head_gross),length(P));

% Estimate flow rate:

for i = 1:length(P_W)
Value = roots([k_a, 0, -Head_gross, (P_W(i)*10^6)/(1000*9.81)]);
for n = 1:3
Qest = (Sn*10^6)/(10000*Head_gross); %Predicting ca. flow rate
if Value(n)<=2*Qest && Value(n)>=0
Qflow(i,1) = Value(n);

135
 References
end
end
end
Q_flow_h = Qflow;
Ploss_Water = k_a*1000*9.81*Qflow.^3*(1/10^6);
Water_N = P_W./(P_W+Ploss_Water);

% Total Plant efficiency and loss:

Ploss_tot = Ploss_Tran + Ploss_Gen + Ploss_Tur + Ploss_Water;
Total_N = P./(P+Ploss_tot);
end

Field current with saturation:

function [Ifd] = Field_current(P, Q, V, Sn, Vn, Ra, Xq, Xd, bv, Xd_tran, X_leak, n_OCC, C_OCC,
Ifd_n)

P = P/Sn; %Convert to Pu values

Q = Q/Sn;
V = V/Vn;
Zb = Vn^2/Sn; %Base impedance
Ra = Ra/Zb; %Pu
I = sqrt(P.^2+Q.^2)./V; %Pu

for n = 1:length(I)
if Q(n) >= 0
phi(n,1) = acos(P(n)./(V(n).*I(n)));
elseif Q(n) < 0
phi(n,1) = -acos(P(n)./(V(n).*I(n)));
end
end
% Unsaturation
delta = atan((Xq.*I.*cos(phi)-Ra*I.*sin(phi))./(V+Ra.*I.*cos(phi)+Xq.*I.*sin(phi)));
Eg = V.*cos(delta) + Ra.*I.*cos(phi+delta) + Xd.*I.*sin(phi+delta);
I_FU = Eg/bv;
% Saturation
Xp = X_leak + 0.63*(Xd_tran-X_leak);
delta_p = atan((Xp.*I.*cos(phi)-Ra.*I.*sin(phi))./(V+Ra.*I.*cos(phi)+Xp.*I.*sin(phi)));
Ep = V.*cos(delta_p) + Ra.*I.*cos(phi+delta_p) + Xp.*I.*sin(phi+delta_p);
IF_p = Ifd_n*(Ep+C_OCC.*Ep.^n_OCC);

I_FS = IF_p - Ep/bv;

Ifd = I_FS + I_FU; %[A]
Ifd = Ifd/1000; %[kA]
end

2D interpolation function for turbine:

function [Tur_N, Hm, Pm] = Turbine(Head_gross, P_Tur, P0_Tur,H0_Tur,N0_Tur)
% A function to ensure no equal numbers due to the function "Griddata"
for i = 1:length(N0_Tur)-1
for k = 1:length(N0_Tur)
if N0_Tur(i) == N0_Tur(k)
N0_Tur(i) = N0_Tur(i)+k*0.000001;
end
end
end
[Hm,Pm] = meshgrid(Head_gross, P_Tur);
Tur_N = griddata(H0_Tur,P0_Tur,N0_Tur,Hm,Pm);
Tur_N = Tur_N/100;
% Alternative method (extrapolation included, but not as accurate within measurements)
%F = scatteredInterpolant(H0_Tur,P0_Tur,N0_Tur,'linear','nearest');
%Tur_N = F(Hm,Pm);

end

136
 References
Function used to collect data from Excel file:
function [P_data, Q_data, V_data, Sys_Price] = Data(data_file)
Q_data0 = data_file(:,2);
P_data0 = data_file(:,1);
P_data = round(P_data0,3); %(MW)
Q_data = round(Q_data0,3); %(MVAR)
Sys_Price = data_file(:,4); %(Nok/MWh)
V_data = data_file(:,3); %(kV)

% In order to eliminate zero-division and unrealistic low average

% values when HPU is turned off. All data points where P = 0 is neglected
m = 1;
for i = 1:length(P_data)
if P_data(m) <= 10
P_data(m) = []; %Here, all values under 10 MW are neglected
Q_data(m) = [];
V_data(m) = [];
Sys_Price(m) = [];
m = m;
else
m = m+1;
end
end
end

137
 References
Function for drawing PQ diagram:
function [] = PQ_diagram(PF,Xq,Xd,fig_n)
% This function is used to draw the PQ (capability) diagram of the
% generator. The PQ diagram consist of Armature limit, Rotor limit,
% Active power limit, Practical stability limit(PSL) and end region heating
% limit/(minimum field current limit).

% This function assumes PSL to intersect the end of active power limit and
% minimum field current limit.

%Nominal values:
Sn = 1;
I = 1;
V = 1;
phi = acos(PF);
k_min = 0.3; % E_qmin = k_min*E_qmax, where k_min is usally between 0-0.3

% Armature limit:
delta_arm = linspace(pi-atan((Sn*cos(phi))/(Sn*sin(phi))), atan((Sn*cos(phi))/(Sn*sin(phi))),
100);
Q_a = Sn*cos(delta_arm) + 0;
P_a = Sn*sin(delta_arm) + 0;

% Active power limit:

Max_Q = [-Sn*sin(phi), Sn*sin(phi)];
Max_P = [Sn*cos(phi), Sn*cos(phi)];

% Field current limit:

% Nominel values
Ax = V^2/Xq;
Bx = V^2/Xd;
delta_n = atan((Sn*cos(phi))/(Sn*sin(phi)+Ax)); % Nominel field limit anlge
En = V*cos(delta_n)+Xd*I.*sin(phi+delta_n); % Nominel excitation voltage

% Maximum Field current liimit:

E_qmax = sqrt((V+I*Xd*sin(phi))^2+(I*Xd*cos(phi))^2);
% Finding the the intersection between Field limit and Armature limit:
delta_field_max0 = 0;
for k = 1:10000
r = E_qmax*V/Xd;
center = -V^2/Xd;
% Simplified method:
Q_fs_max = r*cos(delta_field_max0) + center;
P_fs_max = r*sin(delta_field_max0) + 0;
%P_fs_max =((E_qmax*V)./Xd)*sin(delta_field_max0)+V^2/2*(Xd-
Xq)/(Xd*Xq)*sin(2*delta_field_max0);
%Q_fs_max =((E_qmax*V)./Xd)*cos(delta_field_max0)+V^2*(Xd-
Xq)/(Xd*Xq)*(cos(delta_field_max0)).^2-V^2/Xq;

if (P_fs_max <= (Sn*cos(phi)+0.001) && P_fs_max >= (Sn*cos(phi)-0.001))&& (Q_fs_max <=

(Sn*sin(phi)+0.001) && Q_fs_max >= (Sn*sin(phi)-0.001))
delta_field_max = delta_field_max0;
break
end
delta_field_max0 = k/10000;
end

r = E_qmax*V/Xd;
center = -V^2/Xd;
delta_field_max = linspace(delta_field_max, 0, 100);
Q_fs_max = r*cos(delta_field_max) + center;
P_fs_max = r*sin(delta_field_max) + 0;
%P_fs_max =((E_qmax*V)./Xd)*sin(delta_field_max)+V^2/2*(Xd-Xq)/(Xd*Xq)*sin(2*delta_field_max);
%Q_fs_max =((E_qmax*V)./Xd)*cos(delta_field_max)+V^2*(Xd-
Xq)/(Xd*Xq)*(cos(delta_field_max)).^2-V^2/Xq;

% Finding the practical stability constant "C" and max excitation voltage E
PSL_C = 0;
for C = 0:0.001:1
for E = 0:0.001:2
k = (E.*V)./Xd;

138
 References
a = 1/2*(Xd-Xq)./(Xd*Xq)*V^2; % Constant
d = 2*a; % Constant
Q0 = -V^2/Xq; % Constant
P_PSL = (sqrt(2)*sqrt(k.*sqrt(k.^2+8*d^2)+4*d^2-
k.^2).*(sqrt(k.^2+8*d^2)+3*k))./(16*d); % Practical stability limit (P)
Q_PSL = (k.*sqrt(k.^2+8*d^2)+4*d^2+8*d*Q0-k.^2)/(8*d)+C;
% Practical stability limit (Q)
if round(P_PSL,3) == round(Sn*cos(phi),3) && round(Q_PSL,3) == -(round(Sn*sin(phi),3))
PSL_C = C;
E_max = E;
break
end
end
if PSL_C ~= 0
break
end
end

%Finding the minimum excitation voltage

E_min = 0;
for E = 0:0.001:2
k = (E.*V)./Xd;
a = 1/2*(Xd-Xq)./(Xd*Xq)*V^2;
d = 2*a;
Q0 = -V^2/Xq;
P_PSL_min = (sqrt(2)*sqrt(k.*sqrt(k.^2+8*d^2)+4*d^2-
k.^2).*(sqrt(k.^2+8*d^2)+3*k))./(16*d);
Q_PSL_min = (k.*sqrt(k.^2+8*d^2)+4*d^2+8*d*Q0-k.^2)/(8*d)+PSL_C;
% minimum field current:
for delta_field_min0 = 0:0.0001:pi
E_qmin = k_min*E_qmax; % k_min is usally between 0-0.3
P_fs_min =((E_qmin*V)./Xd)*sin(delta_field_min0)+V^2/2*(Xd-
Xq)/(Xd*Xq)*sin(2*delta_field_min0);
Q_fs_min =((E_qmin*V)./Xd)*cos(delta_field_min0)+V^2*(Xd-
Xq)/(Xd*Xq)*(cos(delta_field_min0)).^2-V^2/Xq;

% Fining when minimum field current is equal to practical stability

% limit
if round(P_PSL_min,3) == round(P_fs_min,3) && round(Q_PSL_min,3) ==
(round(Q_fs_min,3))
E_min = E;
delta_field_min = delta_field_min0;
break
end
end
if E_min ~= 0
break
end
end
% Recalucate minimum field current (with practical stability limit)
E_qmin = 0.3*E_qmax;
d_min = linspace(delta_field_min,0,20);
P_fs_min =((E_qmin*V)./Xd)*sin(d_min)+V^2/2*(Xd-Xq)/(Xd*Xq)*sin(2*d_min);
Q_fs_min =((E_qmin*V)./Xd)*cos(d_min)+V^2*(Xd-Xq)/(Xd*Xq)*(cos(d_min)).^2-V^2/Xq;
%######## Recalculating Practical limit ######
m_PSL = 1;
for E_PSL = linspace(E_min,E_max,20)
k = (E_PSL.*V)./Xd;
a = 1/2*(Xd-Xq)./(Xd*Xq)*V^2;
d = 2*a;
Q0 = -V^2/Xq;
P_PSL(m_PSL) = (sqrt(2)*sqrt(k.*sqrt(k.^2+8*d^2)+4*d^2-
k.^2).*(sqrt(k.^2+8*d^2)+3*k))./(16*d);
Q_PSL(m_PSL) = (k.*sqrt(k.^2+8*d^2)+4*d^2+8*d*Q0-k.^2)/(8*d)+PSL_C;
m_PSL = m_PSL + 1;
end

figure(fig_n)
plot(Q_a,P_a,'Color','r','LineWidth',2.0)
hold('on')
plot(Max_Q,Max_P,'--','Color','r','LineWidth',2.0)
hold('on')
plot(Q_fs_max,P_fs_max,'Color','r','LineWidth',2.0)
hold('on')

139
 References
plot(Q_fs_min,P_fs_min,'Color','r','LineWidth',2.0)
hold('on')
plot(Q_PSL,P_PSL,'Color','r','LineWidth',2.0)
hold('on')
xlabel('Reactive power [Pu]');
ylabel('Active power [Pu]');
grid('on')
hold off
end

The MATLAB program for Åbjøra HPU:

%*********************************************************************
% Author: Sigurd Berg
% Date: 16/05/2021

% This program calculates the efficiency and energy losses

% of a Hydropower unit (HPU), with numerous plots for analysing purposes.

% Credit to "Josè Manuel Amigo" which is the author of the function named:
% densityscatter()
% https://www.mathworks.com/matlabcentral/fileexchange/65024-densityscatter

%*********************************************************************
clear;
clc;

% Read off data points from data file:

data_file = readmatrix('Abjora_OD.xlsx');
[P_data, Q_data, V_data, Sys_Price] = Data(data_file);

% Generator Parapeters:
Sn = 103; %[MVA]
Vn = 11; %[kV]
PF = 0.9; %Power factor
Xq = 0.69; % [Pu]
Xd = 1.06; % [Pu]
Ra = 0.003155; % [ohm] armature winding resistance
Rf = 0.15253; % [ohm] field winding resistance
k_ex = 1.116; % excitation constant (typical value around 1.1)
P_FE = 0.21192; % [MW] Iron loss
P_V = 0.17292; % [MW] Windage and ventilation loss
P_F = 0.2409; % [MW] Friciton loss in bearing and brushes
Xd_tran = 0.15; % Transient D_axis reactance [Pu]
X_leak = 0.11; % Leakage reactance [Pu]
bv = 1.25/694.2; % Air-gap line: Voltage/field current
n_OCC = 9; % Exponent to OCC
C_OCC = 0.12; % Constant to OCC
Ifd_n = 556; % Nominal field current

% Waterway Parameters:
k_a = 0.016335; % Waterway constant
Head_gross = 442.6; % Important: must be within what is given for turbine (only
for advanced turbine method)
%Head_gross = [435, 442, 450];

% Turbine Parameters:

% Simplified method: Turbine_method = 1

% Advanced method : Turbine_method = 2
Turbine_method = 2; %Choose either 1 or 2

if Turbine_method == 1
% Simplified method (independent of head)
Tur_P0 = [16.505 17.571 18.635 19.697 20.758 21.816 22.873 23.927 25.002 26.05
27.096 28.136 29.171 30.204 31.235 32.261 33.285 34.309 35.325 36.338 37.347 38.347

140
 References
39.348 40.344 41.331 42.315 43.298 44.272 45.246 46.216 47.175 48.135 49.099 50.084
51.069 52.058 53.05 54.019 54.989 55.953 56.911 57.859 58.806 59.742 60.678 61.609
62.521 63.435 64.334 65.235 66.121 67.03 67.904 68.781 69.641 70.505 71.351 72.191
73.036 73.862 74.694 75.507 76.313 77.124 77.917 78.709 79.544 80.378 81.21 82.041
82.869 83.695 84.497 85.281 86.049 86.82 87.574 88.321 89.073 89.806 90.532 91.263
91.975 92.681 93.38 94.084 94.793 95.471 96.142 96.806 97.475 98.125 98.767 99.403
100.03 100.65 101.27 101.87 102.45 103.1];
Tur_N0 = [77.406 78.418 79.344 80.196 80.982 81.711 82.389 83.022 83.626 84.18
84.701 85.191 85.654 86.092 86.507 86.901 87.276 87.632 87.971 88.295 88.603 88.897
89.178 89.448 89.704 89.95 90.186 90.412 90.629 90.837 91.036 91.227 91.412 91.591
91.764 91.93 92.091 92.244 92.39 92.531 92.665 92.793 92.915 93.032 93.143 93.25
93.351 93.447 93.538 93.625 93.707 93.786 93.859 93.929 93.995 94.056 94.114 94.168
94.218 94.265 94.309 94.349 94.385 94.418 94.448 94.475 94.498 94.516 94.532 94.544
94.553 94.558 94.561 94.561 94.56 94.556 94.55 94.541 94.53 94.516 94.501 94.482
94.462 94.44 94.416 94.389 94.358 94.328 94.295 94.26 94.222 94.183 94.142 94.099
94.054 94.007 93.959 93.908 93.859 93.797];

Head0 = 0; % must be zero (is used to verify the method)

elseif Turbine_method == 2
%### Advanced method (interpolated between different heads) ###############
% IMPORTANT:
% This method is limited to the active power and head for which are given,
% meaning heads or loads beyond what is given will have no solution and give "NAN"
% Can use the alternative method (commented out) given in the function: turbine()

Head0 = [439,442.6,445.3]; % Vector representing each head (Gross head [m])

%H = 439.2m
Tur_P0(1,:) = [0 16.224 17.274 18.323 19.369 20.414 21.457 22.498 23.537 24.596
25.629 26.66 27.685 28.705 29.723 30.739 31.751 32.76 33.768 34.77 35.769 36.763
37.749 38.735 39.718 40.691 41.661 42.63 43.59 44.551 45.507 46.452 47.399 48.35
49.321 50.292 51.268 52.246 53.202 54.158 55.108 56.054 56.988 57.923 58.846 59.769
60.687 61.587 62.488 63.375 64.264 65.138 66.035 66.897 67.762 68.61 69.463 70.298
71.127 71.96 72.776 73.597 74.399 75.194 75.995 76.777 77.559 78.383 79.205 80.027
80.846 81.664 82.479 83.271 84.045 84.802 85.563 86.307 87.044 87.785 88.508 89.224
89.945 90.648 91.344 92.034 92.728 93.429 94.097 94.759 95.414 96.074 96.715 97.349
97.976 98.596 99.209 99.815 100.41 100.98 101.63];
Tur_N0(1,:) = [0 77.243 78.263 79.197 80.055 80.848 81.582 82.266 82.903 83.513
84.071 84.597 85.091 85.558 85.999 86.418 86.815 87.192 87.552 87.893 88.219 88.53
88.827 89.111 89.382 89.641 89.889 90.127 90.355 90.574 90.784 90.984 91.177 91.364
91.545 91.72 91.888 92.05 92.205 92.353 92.495 92.63 92.76 92.884 93.002 93.115
93.222 93.325 93.422 93.515 93.603 93.686 93.766 93.841 93.912 93.978 94.041 94.1
94.155 94.206 94.254 94.299 94.34 94.377 94.411 94.442 94.47 94.494 94.513 94.53
94.543 94.552 94.559 94.562 94.564 94.563 94.559 94.553 94.545 94.534 94.521 94.506
94.488 94.469 94.447 94.423 94.396 94.366 94.336 94.303 94.269 94.231 94.192 94.152
94.109 94.065 94.018 93.97 93.92 93.87 93.809];
%H = 442.6m
Tur_P0(2,:) = [0 16.505 17.571 18.635 19.697 20.758 21.816 22.873 23.927 25.002
26.05 27.096 28.136 29.171 30.204 31.235 32.261 33.285 34.309 35.325 36.338 37.347
38.347 39.348 40.344 41.331 42.315 43.298 44.272 45.246 46.216 47.175 48.135 49.099
50.084 51.069 52.058 53.05 54.019 54.989 55.953 56.911 57.859 58.806 59.742 60.678
61.609 62.521 63.435 64.334 65.235 66.121 67.03 67.904 68.781 69.641 70.505 71.351
72.191 73.036 73.862 74.694 75.507 76.313 77.124 77.917 78.709 79.544 80.378 81.21
82.041 82.869 83.695 84.497 85.281 86.049 86.82 87.574 88.321 89.073 89.806 90.532
91.263 91.975 92.681 93.38 94.084 94.793 95.471 96.142 96.806 97.475 98.125 98.767
99.403 100.03 100.65 101.27 101.87 102.45 103.1];
Tur_N0(2,:) = [0 77.406 78.418 79.344 80.196 80.982 81.711 82.389 83.022 83.626
84.18 84.701 85.191 85.654 86.092 86.507 86.901 87.276 87.632 87.971 88.295 88.603
88.897 89.178 89.448 89.704 89.95 90.186 90.412 90.629 90.837 91.036 91.227 91.412
91.591 91.764 91.93 92.091 92.244 92.39 92.531 92.665 92.793 92.915 93.032 93.143
93.25 93.351 93.447 93.538 93.625 93.707 93.786 93.859 93.929 93.995 94.056 94.114
94.168 94.218 94.265 94.309 94.349 94.385 94.418 94.448 94.475 94.498 94.516 94.532
94.544 94.553 94.558 94.561 94.561 94.56 94.556 94.55 94.541 94.53 94.516 94.501
94.482 94.462 94.44 94.416 94.389 94.358 94.328 94.295 94.26 94.222 94.183 94.142
94.099 94.054 94.007 93.959 93.908 93.859 93.797];
%H = 445.3m
Tur_P0(3,:) = [0 16.722 17.8 18.876 19.951 21.023 22.094 23.162 24.228 25.315
26.375 27.432 28.484 29.531 30.576 31.617 32.656 33.691 34.726 35.753 36.777 37.797
38.808 39.82 40.828 41.825 42.82 43.814 44.798 45.783 46.763 47.732 48.703 49.678
50.673 51.669 52.669 53.671 54.651 55.63 56.605 57.573 58.531 59.488 60.434 61.38
62.321 63.242 64.166 65.074 65.985 66.88 67.799 68.681 69.567 70.436 71.309 72.164
73.013 73.866 74.701 75.541 76.363 77.177 77.997 78.797 79.598 80.441 81.283 82.123
82.962 83.799 84.633 85.443 86.236 87.011 87.791 88.552 89.307 90.067 90.808 91.541
92.28 93 93.713 94.419 95.13 95.847 96.532 97.21 97.881 98.557 99.213 99.862
100.5 101.14 101.77 102.39 103 103.58 104.24];

141
 References
Tur_N0(3,:) = [0 77.525 78.531 79.452 80.298 81.08 81.804 82.478 83.107 83.707
84.258 84.776 85.264 85.724 86.159 86.572 86.963 87.335 87.69 88.027 88.348 88.654
88.946 89.226 89.493 89.749 89.993 90.227 90.452 90.667 90.873 91.071 91.261 91.444
91.622 91.793 91.959 92.118 92.27 92.415 92.554 92.687 92.814 92.935 93.051 93.162
93.267 93.367 93.462 93.552 93.638 93.72 93.797 93.87 93.939 94.003 94.064 94.121
94.174 94.224 94.27 94.313 94.352 94.388 94.42 94.449 94.476 94.497 94.515 94.53
94.542 94.549 94.554 94.556 94.556 94.555 94.55 94.543 94.535 94.523 94.509 94.494
94.475 94.454 94.432 94.408 94.38 94.349 94.318 94.285 94.25 94.212 94.172 94.131
94.088 94.043 93.996 93.947 93.896 93.846 93.785];
end

% Transoformer Parameters:
P0_Tran = 52.5*10^-3; %(MW) Transformer No-load constant
Pk_Tran = 225*10^-3; %(MW) Transformer loading constant
Sn_Tran = 103; %(MVA) Transformer rating

% Plotting variables
n_var = 100; % Number of parameter variables
Q_n = 20; % Number of Q variables (used in figure 2)

phi0 = acos(0.9);
S = linspace(Sn*0.3,Sn*1.05,n_var);
P = S.*cos(phi0);
Q = S.*sin(phi0);
% P = linspace(Sn*0.3,Sn*0.97,n_var);
% Q = linspace(-20.6,-20.6,n_var);
%Q = linspace(50,50,n_var);
V = linspace(Vn,Vn,n_var);
map_var = [80, 84, 88, 90, 91, 91.5 92, 92.1, 92.185]; % Efficiency lines on efficiency map
%map_var = [0.44, 0.48, 0.5, 0.52, 0.54, 0.56, 0.6, 0.7, 0.8, 0.9, 1]*0.1; % Energy loss
mapping
%map_var = [0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.1, 1.14, 1.16, 1.17, 1.18];

% Data on efficiency
PF_value = 0.9;
phi_value = acos(PF_value);

S_value = 103;
P_value = S_value*cos(phi_value);
Q_value = S_value*sin(phi_value);
% S_value = 103;
% P_value = 0.4*Sn;
% Q_value = -0.21*Sn;
V_value = 11;
Head_gross_value = 442;
[Ploss_Tot,Total_N,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen,Tur_N,Ploss_Tur,Water_N,Ploss_Water] =
HPU(P_value,Q_value,V_value,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv, Xd_tran, X_leak, n_OCC,
C_OCC, Ifd_n, Head_gross_value,k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran);

Ploss_Tran
Ploss_Gen
Ploss_Tur
Ploss_Water
Tran_N;
Gen_N
Tur_N;
Water_N;
Total_HPU_losses = Ploss_Tot
Total_efficiency = Total_N*100
Generator_loss = Ploss_Gen;

%************************************************************************
% PLOTS and other calculations
%************************************************************************

%############################## Efficiency of Components (Figure 1) #################

P_1 = P.';
Q_1 = Q.';
V_1 = V.';
[Ploss_Tot,Total_N,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen,Tur_N,Ploss_Tur,Water_N,Ploss_Water] =
HPU(P_1,Q_1,V_1,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv, Xd_tran, X_leak, n_OCC, C_OCC, Ifd_n,
Head_gross,k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran);

figure(1)

142
 References
f1 = plot(P_1,100*Total_N,P_1,100*Tran_N,P_1,100*Gen_N,P_1,100*Tur_N,P_1,100*Water_N);
xlabel("Load [MW]");
ylabel("Efficiency [%]");
legend('Total HPU','Transformer','Generator','Turbine','Waterway','Location','eastoutside');
set(f1(1),'linewidth',2);
set(f1(2),'linewidth',2);
set(f1(3),'linewidth',2);
set(f1(4),'linewidth',2);
set(f1(5),'linewidth',2);
grid('on')

% It is used later for cumulative probability and efficiency comparison:

Total_N_CP = Total_N;
P_1_CP = P_1;

figure(2)
plot(P_1,100*Total_N,P_1,100*Tur_N);
hold on

% Efficiency at different reactive power

P_eff = P.';
V_eff = V.';
Q00 = zeros(n_var,Q_n);

for m = 1:Q_n
Q00(:,m) = m*(2*Sn)/Q_n-Sn;
Q_eff = Q00(:,m);
[Ploss_Tot,Total_N,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen,Tur_N,Ploss_Tur,Water_N,Ploss_Water]
= HPU(P_eff,Q_eff,V_eff,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv, Xd_tran, X_leak, n_OCC, C_OCC,
Ifd_n, Head_gross,k_a, Tur_P0,Tur_N0,Head0,P0_Tran,Pk_Tran,Sn_Tran);
Total_N0(:,m) = 100*Total_N;
end

Q_s = round(Q00(1,:),1);
figure(3)
for m = 1:Q_n
plot(P_eff ,Total_N0(:,m),'--')
Q_string(m) = sprintf("Q = %3.1f MVAr", Q_s(m));
hold 'on'
end
legend(Q_string)
hold 'off'
xlabel('Active power [MW]')
ylabel('Efficiency [%]')
grid('on')

% % Efficiency vs voltage
% P_volt = P.';
% V_volt = V.';
% Q_volt = Q.';
% [Ploss_Tot,Total_N,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen,Tur_N,Ploss_Tur,Water_N,Ploss_Water] =
HPU(P_volt,Q_volt,V_volt,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,Bv, Xd_tran, X_leak, n0, C, kk,
Head_gross,k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran);
%
% figure(4)
% plot(V_volt,Total_N)

% Efficiency VS Head:
P_head = P.';
V_head = V.';
Q_head = Q.';
Head = 439:0.01:445;
%Head = 300:5:500;
for k = 1:length(Head)
[Ploss_Tot,Total_N,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen,Tur_N,Ploss_Tur,Water_N,Ploss_Water,
Q_flow_h] = HPU(P_head,Q_head,V_head,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv, Xd_tran, X_leak,
n_OCC, C_OCC, Ifd_n, Head(k),k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran);
Total_N_head(:,k) = 100*Total_N;
Turbine_N_head(:,k) = 100*Tur_N;
Waterway_loss_head(:,k) = Ploss_Water;
Q_flow_head(:,k) = Q_flow_h;

143
 References
Turbine_loss_head(:,k) = 100*Tur_N;
end

figure(5)
yyaxis left
plot(Head,Waterway_loss_head(length(P_head),:))
hold on
yyaxis right
plot(Head,Q_flow_head(length(P_head),:))
hold off

figure(6)
plot(Head,Turbine_loss_head(length(P_head),:))

%############## Total HPU efficiency mapping ####################

figure(7)
for h = 1:length(Head)
plot(P_head,Total_N_head(:,h))
B(h) = sprintf("Head %0.1fm", Head(h));
hold 'on'
end
legend(B)
hold 'off'
xlabel 'Active Power [MW]'
ylabel 'Efficiency [%]'
grid 'on'

figure(8)
[Px,Hy] = meshgrid(P_head, Head.');
surf(Px,Hy,Total_N_head.')
xlabel 'Active power [MW]'
ylabel 'Gross head [m]'
zlabel 'Efficiency [%]'
grid 'on'

figure(9)
Nz = round(Total_N_head.',4);
map_var2 = [86:0.5:91, 91:0.4:92, 92.1, 92.1:0.002:93];
[Hc,Hh]=contour(P_head,Head.',Nz,map_var2);
clabel(Hc,Hh)
xlabel 'Active power [MW]'
ylabel 'Gross head [m]'

% Turbine efficiency mapping:

figure(10)
Nz = round(Turbine_N_head.',4);
map_var2 = [94.5:0.01:95];
[Hc,Hh]=contour(P_head,Head.',Nz,map_var2);
clabel(Hc,Hh)
xlabel 'Active power [MW]'
ylabel 'Gross head [m]'

fig_n11 = 11;
Q_n2 = Q_n*5; % Number of Q variables
n_var2 = n_var*5;
P3 = (linspace(Sn*0.1,Sn*0.95,n_var2)).';
V3 = (linspace(Vn,Vn,n_var2)).';
eta_res = 4; % Decimal numbers of rounding of efficiency (should be
equal lim_1 and lim_2)

Q00 = zeros(n_var2,Q_n2);
for m = 1:Q_n2
Q00(:,m) = m*(2*Sn)/Q_n2-Sn;
Q3 = Q00(:,m);
[Ploss_Tot,Total_N,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen,Tur_N,Ploss_Tur,Water_N,Ploss_Water]
= HPU(P3,Q3,V3,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv, Xd_tran, X_leak, n_OCC, C_OCC, Ifd_n,
Head_gross,k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran);
Total_N02(:,m) = 100*Total_N;
%Total_N02(:,m) = Ploss_Tot/Sn %Energy loss mapping
%Total_N02(:,m) = 0.1.*(P3./Ploss_Tot);
end

144
 References
z = Total_N02;
z = round(z,eta_res);
zz = zeros(n_var2,Q_n2);

figure(fig_n11)
[X,Y] = meshgrid(Q00(1,:)/Sn,P3/Sn);
Z = round(z,eta_res);
map_var3 = [88 90 91 91.5 92 92.1 92.2 92.25 92.25:0.05:99];
[c,h]= contour(X,Y,Z,map_var3);
clabel(c,h)
axis([-0.7 0.8 0.3 1])
hold('on')
x1 = [2.5/Sn 2.5/Sn]
y1 = [0 1]
x2 = [-0.6 0.6]
y2 = [81/Sn,81/Sn]
plot(x1,y1,x2,y2)
hold on

%Data point plotting

bins = 30;
Msize = 6;
figure(fig_n11)
densityscatter(Q_data/Sn,P_data/Sn,bins,Msize) % Function created by "Josè Manuel Amigo"
hold ('on')
%################### PQ diagram plotting ####################################
PQ_diagram(PF,Xq,Xd,fig_n11) %Plots PQ diagram
PQ_diagram(PF,Xq,Xd,21) %Plots PQ diagram

% Effieicncy and Power loss of data points

[Ploss_tot,Total_N,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen,Tur_N,Ploss_Tur,Water_N,Ploss_Water] =
HPU(P_data,Q_data,V_data,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv, Xd_tran, X_leak, n_OCC,
C_OCC, Ifd_n, Head_gross,k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran);

Annual_loss = sum(Ploss_tot,'all')
Total_Prod = sum(P_data,'all')
Production_time = Total_Prod/Sn
Average_efficiency = mean(Total_N,'all')

%Efficiency mapping Generator:

fig_n12 = 12;
Q_n2 = Q_n*5; % Number of Q variables
n_var2 = n_var*5;
P3 = (linspace(Sn*0.1,Sn*0.95,n_var2)).';
V3 = (linspace(Vn,Vn,n_var2)).';
eta_res = 4; % Decimal numbers of rounding of efficiency (should be
equal lim_1 and lim_2)

Q00 = zeros(n_var2,Q_n2);
for m = 1:Q_n2
Q00(:,m) = m*(2*Sn)/Q_n2-Sn;
Q3 = Q00(:,m);
[Ploss_Tot,Total_N,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen,Tur_N,Ploss_Tur,Water_N,Ploss_Water]
= HPU(P3,Q3,V3,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv, Xd_tran, X_leak, n_OCC, C_OCC, Ifd_n,
Head_gross,k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran);
Gen_N02(:,m) = 100*Gen_N;
%Total_N02(:,m) = Ploss_Tot/Sn %Energy loss mapping
%Total_N02(:,m) = 0.1.*(P3./Ploss_Tot);
end
z = Gen_N02;
z = round(z,eta_res);
zz = zeros(n_var2,Q_n2);
figure(fig_n12)
[X,Y] = meshgrid(Q00(1,:)/Sn,P3/Sn);
Z = round(z,eta_res);
map_var_gen = [92 93 94 95 96 97 97.5 98 98.2 98.4 98.6 98.8 98.86 98.92 98.96 99 99.1];
[c,h]= contour(X,Y,Z,map_var_gen);
clabel(c,h)
hold('on')

145
 References
PQ_diagram(PF,Xq,Xd,fig_n12) %Plots PQ diagram

% Energy loss mapping HPU:

fig_n13 = 13;
Q_n2 = Q_n*5; % Number of Q variables
n_var2 = n_var*5;
P3 = (linspace(Sn*0.1,Sn*0.95,n_var2)).';
V3 = (linspace(Vn,Vn,n_var2)).';
eta_res = 4; % Decimal numbers of rounding of efficiency (should be
equal lim_1 and lim_2)

Q00 = zeros(n_var2,Q_n2);
for m = 1:Q_n2
Q00(:,m) = m*(2*Sn)/Q_n2-Sn;
Q3 = Q00(:,m);
[Ploss_Tot,Total_N,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen,Tur_N,Ploss_Tur,Water_N,Ploss_Water]
= HPU(P3,Q3,V3,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv, Xd_tran, X_leak, n_OCC, C_OCC, Ifd_n,
Head_gross,k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran);
Ploss_Tot2(:,m) = Ploss_Tot/Sn; %Energy loss mapping

%Total_N02(:,m) = 0.1.*(P3./Ploss_Tot);
end
z = Ploss_Tot2;
z = round(z,eta_res);
zz = zeros(n_var2,Q_n2);

figure(fig_n13)
[X,Y] = meshgrid(Q00(1,:)/Sn,P3/Sn);
Z = round(z,eta_res);
map_var_gen = [0.040:0.002:0.055 0.059, 0.063,0.063:0.008:0.1];
[c,h]= contour(X,Y,Z,map_var_gen);
clabel(c,h)
axis([-0.8 0.8 0.3 1])
hold('on')

PQ_diagram(PF,Xq,Xd,fig_n13) %Plots PQ diagram

% Efficiency VS Voltage:
phi_0V = acos(0.9);
S_0V = (linspace(Sn,Sn,100)).';
P_0V = S_0V*cos(phi_0V);
Q_0V = S_0V*sin(phi_0V);
V_0V = linspace(8,14,100).';
[Ploss_Tot,Total_N_V,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen] =
HPU(P_0V,Q_0V,V_0V,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv, Xd_tran, X_leak, n_OCC, C_OCC,
Ifd_n, Head_gross,k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran);

figure(14)
plot(V_0V,100*Total_N_V)
xlabel 'Terminal Voltage [kV]'
ylabel 'Efficiency [%]'

% Simulaitng with operation data:

% Operation data with zero operation:

data_file2 = readmatrix('Abjora_OD.xlsx');
Q_data2 = data_file2(:,2);
P_data2 = data_file2(:,1);
P_data2 = round(P_data2,3); %(MW)
Q_data2 = round(Q_data2,3); %(MVAR)
Sys_Price2 = data_file2(:,4); %(Nok/MWh)
V_data2 = data_file2(:,3); %(kV)

P_data2(P_data2 < 10) = 0;

P_data_CP = 1*sort(P_data2);

146
 References
P_CP = round(P_data_CP,3)/Sn;
numbers_CP =unique(P_CP);
count_CP=hist(P_CP,numbers_CP);

for i=1:length(count_CP)
if i == 1
prob_CP(i) = count_CP(i)/sum(count_CP);
else
prob_CP(i) = count_CP(i)/sum(count_CP)+ prob_CP(i-1);
end
end
figure(15)
plot(P_1_CP/Sn,100*Total_N_CP) %results from the first figure
hold on
plot(numbers_CP,100*prob_CP)
axis([0.3 1 0 100])
hold on

days = length(P_data2)/(24*4);
t = minutes(0:15:24*4*15*days-15);
[h,m,s] = hms(t);

figure(16)
%yyaxis left
ylabel 'Active power [MW]'
plot(hms(t),P_data2,'R','linewidth',2)
hold on
plot(([1680, 2400]),([72.2, 72.2])) % Average efficiency line
hold on
plot(([1680, 2400]),([77.3, 77.3])) % BEP
xlabel 'Hours [h]'

%#########################################################################

t1 = datetime(2020,1,6,15,0,0);
time = t1 + hours(0:length(P_data)-1);

figure(17)
plot(time,P_data, time,Sys_Price)
xlabel('Date')
ylabel('Power loss [MW]')

P_opt = linspace(76,76,length(P_data)).';
Ploss_opt = P_opt*(1-0.9218);

figure(18)
plot(time, P_data, time, P_opt)

figure(19)
plot(time, Ploss_tot,time, Ploss_opt)
sum(P_data,'all')
sum(P_opt,'all')
%######################################################################

[Ploss_tot,Total_N,Tran_N,Ploss_Tran,Gen_N,Ploss_Gen,Tur_N,Ploss_Tur,Water_N,Ploss_Water] =
HPU(P_data,Q_data,V_data,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv, Xd_tran, X_leak, n_OCC,
C_OCC, Ifd_n, Head_gross,k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran);

average_efficiency = mean(Total_N.','all')*100
average_produciton = mean(P_data.','all')
average_production_r = mean(Q_data.','all')
sum1_prod = (sum(P_data.','all')/4)/10^3
sum1_loss = (sum(Ploss_tot.','all')/4)/10^3

% Dividing the data set into months (manually)

date_name = ['Jan','Feb', 'Mar','Apr','May','Jun','Jul','Aug','Sep','Oct','Nov','Des'];
data_length = [1, 572, 1267, 2011, 2731, 3475, 4195, 4939, 5683, 6402, 7146, 7866, 8610];
for d = 1:12
average_prod(d) = sum(P_data(data_length(d):data_length(d+1)));
average_loss(d) = sum(Ploss_tot(data_length(d):data_length(d+1)));
average_efficiency(d) = 100*mean(Total_N(data_length(d):data_length(d+1)));
end

147
 References
Average_production = average_prod.';
Average_loss = average_loss.';
Average_efficiency = average_efficiency.';

% WAE and EAE calcualtion

Deg = 2; %Number of decimals rounding

des = 100*10^Deg;
P_WAE = P_data;
V_WAE = V_data;
Q_WAE = Q_data;
%################# Collecting and sorting of list ###############
[Ploss_tot,Total_N_WAE] =
HPU(P_WAE,Q_WAE,V_WAE,Sn,Vn,Xq,Xd,Ra,Rf,k_ex,P_FE,P_V,P_F,bv,Xd_tran, X_leak, n_OCC, C_OCC,
Ifd_n, Head_gross,k_a, Tur_P0,Tur_N0,Head0, P0_Tran,Pk_Tran,Sn_Tran);
Total_N_WAE = 100*sort(Total_N_WAE);
N = round(Total_N_WAE,Deg);

N_max = max(N);
size = (100/des):(100/des):N_max;
len = length(size);
val = zeros(1,len); % VALUE of how often each efficiency value
prob = zeros(1,len);

for i = 1:len
m = i/10^Deg;
val(1,i) = sum(N == m, 'all');
if m == 1/10^Deg
prob(1,i) = val(1,i)/length(N);
else
prob(1,i) = (val(1,i)/length(N))+ prob(1,i-1);
end
Ak(1,i) = val(1,i)/length(N);
eta_sum(1,i) = m * Ak(1,i);
end

% Comparion between WAE method and standard average calculation

eta= sum(eta_sum); % WAE
avg = mean(N,'all'); % Standard Average

%line representing average efficiency

x_avg = [avg,avg];
y_avg = [0,100];

val_sum = val(1,len); % initiation (representing maximum efficiency)

for m = (len-1):-1:1 % 1 - 922 or 0%-92.2%
EAE(m) = mean(N((length(N)- val_sum):length(N))); % Expected average efficiency (EAE)
EAE_X(m) = 100 * (val_sum/length(N));
val_sum = val(1,m) + val_sum;
if val_sum >= length(N)
break
end
end
EAE;
EAE(EAE==0)=[];
EAE_X(EAE_X==0)=[];
EAE = flipud(EAE);

figure(20)
h1 =
plot(size(floor(0.01*length(size)):length(size)),100*prob(1,floor(0.01*length(size)):length(si
ze))); % 0.8, means 80% of maximum efficeicny
hold 'on'
h2 = plot(x_avg, y_avg);
hold 'on'
plot(EAE,EAE_X);
xlabel('Efficiency [%]');
ylabel('Percentage [%]');
legend('Cumulative probability', sprintf("Average efficiency = %3.1f", avg));
set(h1(1),'linewidth',2);
axis([91.6 92.23 0 100])
grid('on')
hold on

148