Concept Design of Hybrid Crane Vessel Feasibility Study of Utilizing Electric Energy Storage Technology

Concept Design of Hybrid Crane Vessel Feasibility study of utilizing electric energy storage technology

Zelan Lyu

Concept Design of Hybrid Crane Vessel Feasibility study of utilizing electric energy storage technology

Zelan Lyu

in partial fulfilment of the requirements for the degree of

Master of Science in Marine Engineering

at the Delft University of Technology, to be defended publicly on Tuesday April 19th, 2016 at 14:00 PM.

Report number: SDPO.016.009.m Supervisor: Ir. K. Visser TU Delft Thesis committee: Prof. Ir. J.J. Hopman, TU Delft Ir. K. Visser, TU Delft Dr. W.G. Haije, TU Delft Ing. F. van der Veen, HMC Heerema Ing. R. Wouts, HMC Heerema

An electronic version of this thesis is available at http://repository.tudelft.nl/. Abstract

Heerema Marine Contractors operates large crane vessels, which are exploited for installation works in the offshore oil and gas industry. The benchmark power system of these crane vessels is diesel electric meaning that the main engines are driving alternators for generating electric power. But due to spiky power demand and strict redundancy requirement in dynamic positioning mode of the crane vessel diesel engines on board are exposed to dynamic and extremely low load, which implies very high fuel consumption. Therefore, in order to solve the problem, it is envisaged to investigate the feasibility of hybridization of the benchmark power system, which leads to the research objective of this work:

“Is it technical and economical feasible and advantageous of considering current electric energy storage (EES) technology for improving and optimizing the Thialf on board power generation and distribution system?”

The objective is achieved by making concept design and deriving design criteria for the hybrid power system of the subject vessel by analyzing actual on board measurements. A model of the hybrid power system consists of a battery module, converter module, diesel engine module and control strategy module is created to verify and improve the design, which is used to answer the technical feasibility. An economic analysis is made to quantify fuel saving benefits and capital investment of hybrid system, which is used to answer the economic feasibility.

EES serves three functions on board in the hybrid system: 1) Spinning reserve, replacing one running diesel engine in each engine room and secure power supply during diesel engine failure, 2) Typical peak shaving, absorbing load dynamics (spikes and valleys) so that operation of diesel engine is smoothed in typical dynamic positioning mode, 3) Demanding peak shaving, absorbing load dynamics in the most demanding crane operations.

Results favor Lithium–Titanate battery, flywheel energy storage and supercapacitors in terms of technical feasibility. However, considering dimension and cost analysis as well, Lithium–Titanate battery is selected as the first priority of the hybrid system. The application of designed 446 kWh Lithium–Titanate battery in each engine room resulted in the following energy savings for the operation of subject vessel: 1) Fuel savings by eliminating diesel engine spinning reserve requirement is about 15% per year, which means an estimated yearly 900 k$ fuel saving, 2) Reduction of running hours of diesel engines is about 14,000 hours yearly, which means a saving on maintenance of approx. 300 k$ per year, 3) Furthermore, a fuel and maintenance saving potential by steadier diesel engine operation is showed, but this could not be verified in the model. Therefore, in terms of economic analysis, a yearly benefit of about 1.2 million $ is achieved and considering capital investment is roughly 6 million $ which means the payback period is about 5 years for the hybrid system.

This project revealed some interesting topics for further research. In terms of diesel engine modelling, a model focus on modelling diesel engine dynamic transient operation instead of steady-state performance is of interest. And modelling the diesel engine performance in extremely low loads e.g., 10%-25% is also of interest, although diesel engine is not designed and built to run on such low loads. Recommendations on using a better battery model are also included.

ii | Page

Preface

This master thesis report is the product of last twelve-months-long graduation project carried out at Delft University of Technology and Heerema Marine Contractors (Leiden). The idea of the topic origins from Heerema Marine Contractors Leiden office, Asset Management Department, Equipment Support Group. And finally, in this work, it proves that the idea is not only a pioneering and interesting one in oil & gas industry but also a smart and feasible one. I would very like to express my thanks to the following people:

From TU Delft:

• Klass Visser, who is my supervisor at TU, for his throughout guidance and supervision, valuable questions and excellent advices, which enlightened me throughout my graduation. • Hans Hopman, for his supervision, efficient feedback and interest he showed during the meetings and being the chairman of the committee. • Wim Haije, for being the committee member, very finely reviewing of my report and gave constructive criticism. • Peter de Vos, who introduced this project to me one year ago so that I didn’t miss the opportunity for such an interesting and challenging project.

From Heerema Marine Contractors:

• René Wouts, who is my daily supervisor at Heerema and has always been supportive and enthusiastic, for his attitude, knowledge and experience he shared with me, which set the example of a good engineer for me. • Fokke van der Veen, who is the leader of Equipment Support Group. His communication skills, encouragements and suggestions during my project were admirable. • All the colleges at Asset Management Department especially Equipment Support Group, this page is too short to name all of you, that made my time at Heerema a very pleasant one.

Furthermore, I would like to thank:

• Hong Zhou, Sotiris Kouroutzis and Rinze Geertsma for discussing diesel engine A4 model with me. • Valentin Muenzel, for discussing his Lithium-ion battery cycle life paper and model with me. • Louis-A. Dessaint and Olivier Tremblay for answering my questions on battery model in SimPowerSystems toolbox.

Last but most important:

I’d like to thank my parents and family for always believing and supporting me including paid a huge amount of tuition fee to support my master study at the Netherlands. Definitely, my friends, for sharing the joy, the interest and the beer with me as well. My girlfriend, for being my mirror, helping me and encouraging me all the time.

Zelan Lyu Delft, April 2016

iii | Page

“The main issue in this discipline is “system integration”: the integration of different equipment and disciplines to create well-functioning, efficient and cost- effective systems.”

— Delft University of Technology Study Guide-Marine Technology-Design, Production and Operation-Marine Engineering

iv | Page

Contents 1. Introduction ...... 1 1.1 Background...... 1 1.2 Review of EES technology developments in marine sector ...... 4 1.3 Synthesis...... 7 2. Research objective ...... 8 2.1 Problem definition ...... 8 2.2 Research objective ...... 12 2.3 EES functions and benefits on Thialf application ...... 12 2.4 Hybrid system challenges and disadvantages ...... 13 2.5 Main research question and outline of the thesis ...... 14 2.6 Research scope ...... 17 2.7 Synthesis...... 18 3. Functional decomposition ...... 19 3.1 Thialf functions ...... 19 3.2 System & component ...... 21 3.3 Energy flow diagram (EFD) ...... 25 3.4 Synthesis...... 27 4. Operational profile ...... 28 4.1 General ...... 28 4.2 Load data analysis ...... 32 4.3 Fuel saving potentials ...... 41 4.4 Synthesis...... 43 5. Electric energy storage (EES) ...... 43 5.1 Configuration ...... 43 5.2 Ragone plot ...... 51 5.3 Alternatives ...... 53 5.4 Synthesis...... 63 6. Modelling ...... 63 6.1 Goal of modelling ...... 63 6.2 System integration and overview ...... 64 6.3 Diesel engine model ...... 64 6.4 Converter model ...... 71 6.5 Battery model ...... 72 6.6 Control strategies ...... 85 6.7 Synthesis ...... 89 7. Economic analysis...... 90 7.1 Benefits ...... 90 7.2 Cost ...... 90 7.3 Synthesis...... 92 8. Conclusions & recommendations ...... 92 8.1 Conclusions ...... 93 8.2 Innovative work ...... 97 8.3 Recommendations ...... 97 9. Discussions ...... 99 9.1 Topic 1: Safety ...... 99 9.2 Topic 2: Fuel cell ...... 99 Bibliography ...... 100 APPENDICES ...... 104 A. EES energy capacity (for typical peak shaving function) ...... 104 B. Peak analysis ...... 106 C. Overview technical data for EES ...... 107 D. Estimation of practical values for SC ...... 109 E. Estimation of practical values for FES ...... 109 F. Generator efficiency ...... 112 G. Results comparison (Excel & DE A4 static) ...... 113 H. High power Li-ion batteries ...... 115 I. List of abbreviation...... 115

vi | Page

1. Introduction

Electric energy storage (EES) technology have been focusing on storing energy for electrical purpose since the widespread introduction of electricity. Early examples are pumped hydro storage on large scale due to its geographic features and batteries on small scale due to their relatively small capacity and high cost.

Nowadays, a plethora of competitive EES technologies are becoming increasingly critical and growing needs in the drive to increase the efficiency and effectiveness of different power systems. For example, renewable energy sources such as wind and solar electricity generation farms require energy storage to buffer power production deficits. Interesting approaches have been applied in warehouse-scale datacenters using EES to either shift demand away from high tariff periods, or to shave demand allowing aggressive under-provisioning of the power infrastructure. In ground transportation market, EES like advanced batteries which have been showing a promising development in both cost and energy density, afford unprecedented opportunities for hybrid cars which charge using low-cost energy from the grid and recover the energy otherwise dissipated during braking like Toyota Prius and even full electrical cars like Tesla.

Marine sector thereby has been the beneficiary of these industries that have accelerated EES technology. Although the primary storage on ships is in fuel and the energy stored in fuel was used nearly exclusive for propulsion in most instances today. But with the growth of large pulsed loads on vessels which have a variable operational profile e.g. crane vessels, offshore supply vessels (OSV), ferries, fishing vessels and especially navy ships with large electric consumers like radars, electric weapon systems and etc. the situation is changing. There is an increasing opportunity for additional EES technologies to play a role. Although these loads can, in principle be operated successfully directly from fuel, it appears that the power system may be smaller, lighter, more efficient, more reliable and more environmental friendly in some specific instances with additional EES technologies [1].

1.1 Background

In marine sector, EES units are already installed on board to ensure continuous power supply to important loads such as operating, monitoring and control systems during startup of the emergency system after a blackout of the primary supply system [2]. Likewise, in applications where the EES system is the main energy source, which is also relatively straightforward since it can be treated much like a traditional diesel generator (DG). However, in cases where its purpose it to enhance the overall performance of a system, and do so in parallel with DG, this requires more attention. In such systems the EES device can take on a wide range of functions and purposes (or advantages). Six potential functions are respectively defined below with corresponding advantages identified as well. It should be noted that the terminologies define different EES functions used in this work are quoted from a published report of ABB1 [3] . In other different commercial and academic reports, different terminologies might be used for the same EES function.

1. Spinning reserve2. EES device is connected and running but not charging or discharging energy into the system. On loss of generating capacity it steps in to take the load for a predefined period of time. If other functions are activated simultaneously, this function ensures that sufficient energy is left in EES device.

a. Backup for running DG.

b. Fewer engines needed online.

c. Improved fuel efficiency.

d. Reduced engine running hours.

2. Enhanced ride through. Same as spinning reserve, but in a local level in a sub- system like a thruster or crane drive.

a. EES device can give UPS (uninterruptible power supplies) like functionality for all or portions of power system.

3. Peak shaving. Unit shaves short duration peak power and absorbs load variations in the network so that engines only see the average system power.

a. Level the power seen by engines.

b. Offset the need to start new engine or use smaller engine.

c. Improved fuel efficiency.

d. Reduced engine running hours.

4. Enhanced dynamic performance.3 Unit absorbs sudden load changes and then ramps the changeover on running engines. If peak shaving is used, then this function is automatically included.

1 ABB is a company specialized in power and automation technologies. 2 Spinning is derived from hydroelectric and combustion turbine terminology. Reserve generator turbines can literally be kept spinning without producing any energy as a way to reduce the length of time required to bring them online when needed. 3 In this work, the author makes no difference between the terminologies “peak shaving” and “enhanced dynamic performance”. But in the strict sense, no matter which terminology the author used in this work, it actually refers to the “enhanced dynamic performance” according to the definition of ABB.

2 | Page

a. Instant power in support of running DG.

b. Enable use of slower engines like LNG (liquefied natural gas)/dual fuel engine and fuel cells.

5. Strategic loading. Unit charges and discharges to optimize the operational point of running engines, ensuring that energy is produced at the lowest cost, taking the efficiency of the EES system into account.

a. Charging and discharging EES media in such a way that it optimizes the operating point of the DG.

b. Power is produced at peak efficiency of DG.

6. Zero emissions operation. Unit powers the system so that engines can be turned off.

a. Zero emissions in harbor.

b. Quiet engine room.

There are also other fundamental compelling reasons for some degree of EES like the fuel-to-electricity storage option is not reversible. Reversibility can be important for efficient operation in some cases. For example, in electric guns on navy ships, when the projectile leaves the barrel, considerable energy is stored in the collapsing magnetic field between the rails. In a reversible system, this energy can be captured, stored, and used in the next shot [1]. In a crane vessel energy can also be stored when the crane is lowering load and feedback to the grids during peak demand periods. This idea has already been applied in the Rubber Tired Gantry (RTG) cranes loading and unloading containers in ports like Rotterdam considering the high cycling characteristics of these cranes.

In conclusion, by integrating EES device in the power system in these ways, it becomes possible to run the power system with fewer engines online, thereby reducing accumulated running hours of the power plant, which leads to reduction of maintenance workload. Running fewer engines also means increasing their partial loading which in turn improves specific fuel consumption [3] and reduces low-load induced maintenance4. What’s more, engine transient load variations can be largely reduced as well, in other words, engine runs much steadier that improves combustion process in the cylinder, which lead to extension of lifecycles, reduction of maintenance work and also potential fuel savings. Last but not least, more efficient and steadier operation of diesel engine can reduce emissions as well. Thus, a hybrid power system can be more efficient, environmental friendly and competitive than a conventional one if the right EES

4 Low load maintenance is for instance internal cleaning of the engine due to pollution which occurs at low load. Alternatively the engine can be regularly operated at high load so that the engine is “cleaned” by itself (high temperature).

3 | Page technology can be perfectly integrated with conventional power system.

1.2 Review of EES technology developments in marine sector

EES technologies have already been used in marine industries with a long history, for instance most 20th-century submarine used batteries for propulsion when submerged and diesel engines on the surface, and for battery recharging until the advent of nuclear propulsion. Recently, interest approaches have been applied in ships like the Viking Lady—world’s first hybrid offshore supply vessel developed by Wartsila, DNV and her Norwegian ship-owner [4], Ampere—the world’s first large all-electric (100% battery) driven aluminum catamaran ferry in Norway [5], the first two sea-worthy hybrid ferries in Scotland [6], etc.

1.2.1 Hybrid platform support vessel (PSV) The Viking Lady is a state-of-the-art 92m PSV delivered in 2009 and classed by DNV. She was especially designed to safely service offshore installations in the extremely harsh waters of the North Sea but what makes this lady really unique is her hybrid power system, including four LNG-powered Wartsila 32DF dual-fuel5 driven generator sets of which each has a capacity of 1950 kW, an energy storage system in the form of a 442 kWh demonstration battery package (full scale expected to achieve 2-3 MWh) and a 330 kW high temperature fuel cell energy source specially adapted for marine use. What’s more, new class rules for Battery Power and updated rules to cover Hybrid Energy System have been developed by DNV in 2011 and 2012 respectively as a part of the project [4, 7, 8]. The Viking Lady is a milestone of EES developments in marine sector as a product of the FellowSHIP project [7] starts form 2003 since she is the first ship commercially use fuel cells and first hybrid PSV.

She’s fitful and required to do different tasks all the time that will see engine load swing from 80% down to as low as 10%. Thus, the variable operational profile enables a huge potential for fuel saving, which is the basic object to install the hybrid system onboard. Viking lady test installations run in several modes:

 Island Mode: Fuel cell and battery alone power the vessel.

 Transit Mode: The normal operational philosophy is to have some redundancy in this mode in order to manage large instant power peaks due to irregular weather conditions and the onboard utility systems which means that one extra engine is in

5 Wartsila multi-fuel engine can run in the following operation modes, and switch between them seamlessly without interrupting the power supply: • Gas only (with liquid pilot fuel): natural gas, LNG, biogas, associated gas. • Liquid fuel only: Crude oil, diesel, residual oil, fuel-water emulsions, liquid biofuel. • Fuel sharing mode: Gas and liquid fuel simultaneously.

4 | Page

operation. This will be avoided by using batteries to reduce fuel consumption. What’s more, Viking Lady is en route (particularly for heavy weather condition) where the battery improves response time of power system and reduces engine dynamic transient load variations (peak power shaving) to smoother the engine operation.

 DP (dynamic positioning) Mode: Battery is used as power redundancy in case a fault condition occurs as well as adopting a charge/discharge strategy allowing the engines to run at more efficient loads in addition to reducing frequent load variations, otherwise, two engines must run at low loads to ensure redundancy. (The testing has been performed as a normal DP operation outside the critical operational zone.)

Future step for Viking Lady is to only use battery as power source in harbor to stop the engines. In essence, a totally clean and remarkably silent ship would shuttle into port. And it is also expected in the future to use battery as a backup for engines e.g. Viking Lady maybe run on one engine when it is alongside platforms. Then the load can be increased on the remaining engine and thus system efficiency is improved. If anything goes wrong battery can be backup until the start of standby engine.

The fuel cell and combustion engine on board both powered by LNG which already significantly reduced emissions and fuel cell additionally leads to reduced fuel consumption (up to 30% increase in energy efficiency), reduction in requirements for maintenance, safer operation and quiet operation. Fuel consumption, emissions and maintenance thus should be declined as a result of a more efficient (optimal load) and smoother operation of the engines and using batteries as a redundant source of power in hybrid system. During more than a year of monitoring, measuring and documenting fuel use, emissions and a wide range of other factors in operation of the Lady before and after installing the hybrid system, the team found that the new system has cut the fuel needs by 15%, which means the new system can pay for itself within a few years [4, 8].

1.2.2 Hybrid ferry MV Hallaig and MV Lochinvar are the only two sea-going passenger and vehicle roll-on, roll-off ferries in the world to incorporate a hybrid system of diesel electric and lithium-ion batteries power, which are commissioned and owned by Caledonian Maritime Assets Ltd (CMAL) in Scotland.

Both ferries are 43.5m long, accommodating 150 passengers, 23 cars or two heavy good vehicles. What’s special about the hybrid system is that the two 400 kW, 400 Volt permanent magnet motors can be powered in different ways since there are three 330 kVA diesel-generators and two 350 kWh lithium-ion battery containers on each ferry. Converters ensure that the direct current provided by batteries can be transformed into 400-Volt alternating current. In principle one generator can power the ship. Using batteries first when extra energy is required (for manoeuvers or in case of bad weather) instead of an extra generator can save 19-24% on fuel over a conventional diesel mechanical solution. Moreover, the batteries can power all the electricity needed on board when the ship is in the harbor [9]. The battery banks will be charged overnight from the mains. It is also worth noting that a much higher degree of redundancy on board is

5 | Page achieved on the hybrid ferry compared to the previous designs. An extreme situation is if all power from diesel electric generators are lost the ship can still run on batteries alone, thus, on days with a reduced number of crossings between islands the ferry can be operated only on batteries.

In addition, the beauty of the system is that not only do the batteries alone reduce fuel consumption. The Energy Management System (EMS) (from Imtech Marine) very cleverly optimizes the load sharing between the lithium ion batteries and generator whilst in hybrid mode, allowing the generator to run as efficient as possible. The EMS and the whole optimization process are the key to these fuel savings. An even up to 38% fuel savings compared to the baseline of operating in diesel electrical mode without batteries on the MV HALLAIG is recorded during trails. The bulk – some 28% - came from shore charging the batteries overnight. This is energy that doesn’t have to be generated during the day by the diesel generators. The remaining 10% of additional fuel savings is the result of the fact that batteries are used in a “smart” way via EMS [10, 11].

1.2.3 Full electric ferry using battery Ampere—the world’s first large all-electric car and passenger ferry has been in operation since early 2015. The ferry features an all-electric power train, with two electric motors with 450 kW of output each. The 80-meter-long vessel is able to carry 120 cars and 360 passengers across the Sognefjord, in Norway [5]. The double-ended ferry plies its 5.6km single route 34 times daily at an average speed of 10 knots, with a crossing time of 20 minutes using 150 kWh energy. Pausing time at quay is 10 minutes, which is used to charge the on-board, 1-MWh lithium-polymer battery pack (two modules mounted on each end of the ferry) [5]. Another two shore power stations of 410kWh battery package each are installed on each end of the route to serve as a buffer. The battery’s low impedance and advanced liquid cooling system, combined with Siemens high-power charging equipment, supports 2.5 times the continuous current of the vessel-mounted batteries which facilitates this rapid battery-to-battery power transmission. And once the ferry has left, the shore-side battery slowly recoups energy from the local medium- voltage grid (powered by hydroelectric plant) at a rate permitted by the grid infrastructure based on other demands, which reduce the need for expensive upgrades to electrical grid infrastructure at the ports [12]. Across a 12-month period, it’s thought that the emissions-free ferry will consume around two million kilowatt-hours of energy, whereas conventional diesel ferries consume at least one million liters of fuel and emit around 570 tons of carbon dioxide and 15 tons of nitrogen oxides [13].

1.2.4 Full electric ferry using supercapacitor In 2013, STX France Lorient has delivered to Lorient Agglomeration the world’s first full electric supercapacitor 22m catamaran, Ar Vag Tredan. She plys a 20 minutes round trip of around 1.7 nautical miles 28 times per day at a maximum speed of 10 knots crossing the Lorient bay between the city centre and Pen Mane, where she recharges her supercapacitors in just four minutes by industrial plugs (capacity up to 3.3kV AC) with voltage of 400V AC 400 A. 128 supercapacitor modules with a total weight of 6 tons are distributed in the two hulls and have a total capacity of around 28 kWh, which is more

6 | Page than enough for a round trip at service speed. Two 100 kVA gensets are also installed as back-up power in case of grid failure at charging point and bad weather conditions.

The main challenge of using supercapacitors is to control the discharge rate to make sure that the supercapacitors do not release their energy too quickly, but more slowly over a longer duration of 25 minutes [14]. The software-based energy management system handles this task and forms the core of the supercapacitor solution. And the system automatically switches on the diesel generators when there is only 5 percent of electric energy left in supercapacitors [14].

The advantage of supercapacitor compared to batteries is that no maintenance is needed during the long last millions of life cycles. Other safety reasons like no fire or explosion risk, no special handling conditions, no gas emissions, no sensitivity to a full discharge, short circuit or fire, which leads to the final installation of Bureau Veritas-certified supercapacitors on board. Furthermore, the charge-discharge efficiency of supercapacitors is 98%, as there is no power limitation for this technology. The only limitation is at the converter level (maximum current when discharging)[14-16].

1.3 Synthesis

There are still many other interesting developments of EES technology applications on board not mentioned above like the TESO ferry, which is about to start construction in Spain, mixes LNG, fuel oil, batteries and even solar power [17]; today six hybrid ferries operated by Scandlines now comprise the largest hybrid fleet commercially in operation. But propulsion is not the only area that could benefit from hybrid solutions – there’s also the draw of hotel loads and onboard equipment. Japan’s Mitsui OSK Lines car carrier, the Emerald Ace, delivered in 2012, retains its conventional two-stroke propulsion but solar panels take over electrical generation en route, tapping into stored battery energy while at berth [18]. Grieg Star and DNV recently collaborated to simulate crane operations on one of Grieg Star’s open hatch vessels. DNV’s COSSMOS simulation tool was used to model and compare the use of onboard diesel-electric gensets to power four cranes, and the use of battery hybrid sets. The results were startling, as the hybrid ship used around 30% less fuel and the equipment had a payback time of less than a year [19].

Thus, from the above review of recent latest developments of EES systems in marine sector, the degree of on board EES applications have been explored in depth not only on the capacity of the storage but on how to use the latest EES technology [17]. The driver behind EES applications in maritime industry has been the desire to efficiently utilize potentials of onboard modern electric power supply system with respect to operational costs, emissions and flexibility. It is also the point of this work, which is to facilitate a mix of DE and EES so that a breadth of systems are available to choose from depending on what is needed in terms of operation whilst considering technical challenges and economic returns.

7 | Page

2. Research objective

In this chapter research objective of this work is defined: in sub-chapter 2.1, it is studied and analyzed what the main characteristic and problem of current/benchmark power system is. In sub-chapter 2.2, the research objective is proposed in order to solve the problem. In sub-chapter 2.3, technical functions of EES including potential benefits are researched in detail. In sub-chapter 2.4, technical challenges and disadvantages of integrating the EES into the existing power system are summarized. Maine research objective is decomposed into sub-questions in sub-chapter 2.5. The research scope is summarized in sub-chapter 2.6 to interpret why the main problem defined in sub-chapter 2.1 is tackled in the proposed angle.

2.1 Problem definition

Heerema Marine Contractors (HMC) operates large crane vessels, which are exploited for installation and decomposition works in the offshore oil and gas industry. SSCV (semi- submersible crane vessel) Thialf, target of this project, is one of its semi-submersible crane vessels and classified as a DCV (deep-water construction vessel). She has been equipped with two cranes that offer load lifting capacitance of 14,200 tons at 31.2 meters, which makes her the biggest crane vessel in the world. A maximum weight of about 12,000 tons can be loaded onto the working deck, in similar terms, 15 tons per square meter of the deck. She has a gross weight tonnage of 136,709 tones with a net tonnage of 41,012 tons and with a length, beam and maximum draft of respectively 201.6m (this is overall length; length of vessel is 165.3m), 88.4m and 31.6m she can easily accommodate a crew of over 700 members. Transit speed with 12000 t deck load is 6 knots at 12.5 m draft and with 1 tug maximum 7 knot at 12.5 m draft. Thialf is equipped with a Class 3 DP system with six 360 degrees 5,500kW azimuth thrusters providing 420 tons total thrust [20]. Figure 2-1 shows installing a topside by Thialf.

8 | Page

Figure 2-1 Thialf (Source: HMC)

2.1.1 Power configuration The power generation of these crane vessels (including Thialf) is diesel electric meaning that diesel engines which drive alternators for generating electric power are fuelled using marine diesel oil (MDO). The 12 diesel generators installed on Thialf are fully separated into three independent engine rooms and have a total nominal capability of 56.4MW power providing for different operations. The electric power is then distributed throughout the vessel by power management and distribution system for powering all electric consumers. The power consumption can be divided in three distinct groups:

• Hotel load • Thrusters • Cranes

2.1.2 Hotel load The hotel load is relatively constant of nature and relatively low when compared to the power generation capacity. The hotel load comprises the electric power that is required for the lights, the heating, ventilation and air conditioning systems (HVAC), the laundry, the mess rooms, various hydraulic systems, air compressors and etc.

2.1.3 Propulsion load The thrusters are propeller units which are driven by large electric motors. These thrusters are used for sailing the vessel to the next offshore installation location, or for keeping the vessel at the envisaged position as part of the DP system. The electric power that is consumed by all the thrusters depends on the type of operation and on the experienced environmental conditions. Normally the total consumed power varies gradually in time but sometimes it can change significantly fast and become as high as the maximum capacity of the diesel generators. Protection systems should limit the thruster power consumption in order to prevent potential overloading of the diesel generators.

2.1.4 Crane load There are two main cranes with main hoist, auxiliary hoist and whip hoist installed on

9 | Page

Thialf. Cranes are electrically driven by DC (direct current) motors. Power consumption of the cranes is very irregular of nature. Power consumption can ramp up very quickly to relatively high power levels when the crane is hoisting heavy loads. And a relatively high and short power demand is also required for initiating the movement of the crane, but almost no power is required for maintaining the initiated movement. Thus, sufficient diesel generators should be online at all times to manage the fast and high power demand. Except heavy lifting operations like installing topside on the jacket, actually most of the time the cranes are operated for daily relatively light lift (or empty crane) operations like relocating different cargos or equipment on the deck and loading light cargos like coffee from barges to the deck. However, around 2 MW power is required for rotating the main crane, therefore, the power demand is then very spiky of nature. The peak power demand of the crane requires a surplus of power generation capacity, which implies additional running DG(s) to manage the fast and high power demand. This is also the main reason for having adopted power limits for the propulsion load.

Last but not least, when lowering the load, motors of the crane are used as brakes, actually becoming generators that create free regenerated energy, which could be put to use but is currently directed to onboard resistor banks and wasted as heat on Thialf.

2.1.5 Operational profile In terms of Thialf which has been operated for about 30 years, recorded operational profile contains evidence of problems. Understanding the problem indicates the research direction and also provides inspiration for potential solutions. Figure 2-2 offers an observation according to recorded data on last five years, which shows variable operational modes of Thialf, divided into the following: Standby Inshore, Transit, Standby DP-1, and DP-2/3 Operation (heavy lifting operation is based on DP mode). To anthropomorphize a bit, she’s fitful.

Operational Modes

15% Standby Inshore

Transit 47% Standby DP-1 32% DP-2/3 Operation

Figure 2-2 Thialf operational modes

Thialf is dominant by DP-2/3 operation around 50% of time, which involves strict requirements regarding power redundancy where life and great economic values would be endangered by a single diesel engine failure. Normally two or more engines in one engine room have to be in operation to secure back-up power for the required load in case a fault condition happens. Figure 2-3 with more details shows a nearly 17 hours’

10 | Page recorded load consumption in time domain of starboard (SB) engine room within Thialf during DP-2/3 operation.

Figure 2-3 Load consumption of SB engine room during DP operation6

Three DGs (NO. 2, 4 and 6) are running to provide a total power of 13.8 MW as shown by the green dash line in Figure 2-3 during DP operation. It can be concluded from the figure that Thialf applies an extremely conservative strategy to deal with the irregular instant pulsed load due to irregular weather conditions and crane operations. More specifically, one DG can fulfill the load consumption most of the time but in case of bad weather and due to redundancy requirements another DG is brought on line. Heavy lifting operation based on DP mode implies one more DG should be online for irregular short spiky lifting load, therefore, in total three diesel engines (DE) are sharing loads and running under extremely low load of less than 20% most of times. Figure 2-4 shows the specific fuel consumption (sfc) of DE 7 and 8 on Thialf. Fuel consumption increases almost nonlinear with the decrease of load below optimal load point around 75% to 80%.

Figure 2-4 Specific fuel consumption of 8ZAL40S (at constant speed of 510 rpm at each point)

Therefore it can be concluded that the benchmark power system onboard is fully energy

6 It is assumed to be a typical operational profile of Thialf on DP operation.

11 | Page inefficient, which is mainly due to: • Low load operation of DE. Additionally, it is assumed that another potential reason of inefficient DE operation is: • Dynamic transient operation [21, 22] (please refer to chapter 6.3.2).

2.2 Research objective

According to chapter 2.1 the potential fuel saving of benchmark power system is high. For example, hybridization is an option - replacing one DG in each engine room by EES device. The EES device like battery can be used in a charging/discharging strategy for peak load shaving and redundancy operations like providing a back-up safety margin when one generator shuts off for some reason, it will provide power throughout the engine room for at most 5 minutes7 with the other DG until the standby generator is brought online.

Therefore within hybrid power systems, DE(s) can be loaded steadier, more efficient and nearer to their optimum operational point. The benefit of more efficient and steadier DE operation is fuel savings as more sufficient burning of fuel and much less wear and tear due to load variations. What’s more, despite fuel savings, running one engine less in each engine room means less running time of DG sets and also less wear and tear which results in less maintenance workload. Therefore, it is a good objective to install an EES system on board. Although the spinning reserve mandatory under DP2 and DP3 rules has until now only been answerable by engine power, it seems the classification societies and developers are at present involved in proving that battery packs could take over this traditional role, with the innovative Viking Lady commencing trials with batteries to demonstrate the technology [18].

It should be noticed that potential technical challenges and disadvantages such as system complexity, huge capital investment, etc. also come with benefits of the hybrid power system.

Therefore, to sum up, the main research objective of this thesis is:

2.3 EES functions and benefits on Thialf application

For Thialf, which is the primary scope of this research (refer to 2.6), since two or three DEs are sharing load and thereby running on extremely low load in three engine rooms. Therefore, as a most important step, using EES as spinning reserve to replace one running DG in each engine room, which leads to the hybridization of exist power system.

7 The standby DG can be started faster but the worst situation is 5 minutes.

12 | Page

Several benefits thereby can be achieved like reduction of accumulated running hours of DGs, which leads to reduction of maintenance workload. Running fewer engines also means increasing their partial loading which in turn improves specific fuel consumption and reduces low-load induced maintenance.

Secondly, periodically quick load transients must influence the fuel consumption differently according to the type of engines, which can be handled by the EES system. Therefore, it is envisaged to conceptually develop the EES system which could also be able to deal with the fast and high power demand during spiky loads. Following strategies like charging/discharging can then also be implemented to further reduce engine load variations. Charging can be done using the on board diesel generator sets when load is lower than basic load.

To sum up, the potential of EES on Thialf will be investigated to bridge power and energy demand during excessive power demand or main engine failures, which should in principle provide following technical functions:

1. Spinning reserve—replace one engine for backup power in DP operation allowing predefined period of at most 5 minutes for load dependent start of additional diesel generator sets for fulfilling consumed loads.

2. Enhanced dynamic performance— continuous peak shaving of pulsed loads and absorbs load steps to ramp-up/-down the engines gently in order to prevent or reduce big load fluctuations for running diesel generator sets. (Although the engines end up taking the load anyway, but how it is presented has a huge impact on maintenance and life, engines don’t like to be stressed.)

3. Store the regenerative energy during lowering loads or de-accelerating motions of cranes.

Thus, various potential advantages will supposed to be achieved based on a steadier and more efficient load of the DG sets as follows:

1. One less DG online for each engine room. 2. Less maintenance workload. 3. Less fuel consumption. 4. Less emissions. 5. Potential use of cleaner fuel (LNG) with inherent slower engine response for future design.

2.4 Hybrid system challenges and disadvantages

It should be realized that hybridization of Thialf onboard power system implies not only benefits but also challenges and disadvantages as briefly mentioned in chapter 2.2. The most important two aspects considered in this work are system complexity and investment cost.

13 | Page

System complexity is defined in two dimensions: hardware and software. Hardware indicates how to integrate EES onboard. Firstly, the location of EES in the power system (in a centralized level or decentralized level) is determined in chapter 3. Secondly, design criteria for each EES specification e.g. energy capacity, power capacity, etc. can be defined according to its three functions, which is defined in chapter 4. Thirdly, which EES technology can be the optimal alternative for Thialf application is determined in chapter 5. Additionally, the physical topology of integrating EES into the onboard high voltage (HV) alternating current (AC) grid is also briefly mentioned in chapter 5. Software indicates how to integrate EES into the existing PMS system e.g. how to charge the EES device is interpreted in chapter 4.2, how to design the control strategy is illustrated in chapter 4.2.2 and which control strategy may be favored is determined in chapter 6.6.

Investment cost is always of concern, and sometimes is of overriding importance, even at the expense of performance. Therefore investment cost of the system is investigated and compared with the benefits (refer to chapter 7). Payback period thereby can also be estimated, which provides valuable information for decision makers.

2.5 Main research question and outline of the thesis

2.5.1 Main research question The abovementioned considerations lead to the formulation of a main research question and its related sub-research questions. The main research question is:

Is it technical and economical feasible and advantageous of considering current electric energy storage (EES) technology for improving and optimizing the Thialf on board power generation and distribution system?

This question can only be answered when the following questions are answered:

1) Functional decomposition and energy flow diagram.

Answering this question provides the vessel’s main goal, the mission, into the main functions that need to be performed to achieve this goal. The functional approach forces the author to consider why the EES and other systems need to be on board, instead of how they are split up in components (i.e. the big picture vs. details). The coherence and interaction between systems (but also within the system, between components) would be more difficult to comprehend, if the function of the system is not the starting point of the analysis [2]. The ship’s machinery systems require energy of different forms: for instance fuel, electric power and mechanical power. Energy flow diagram (EFD) models the chain of energy conversions from energy source to the types of energy required for the systems: the energy use, thus, it shows the connection between systems. It is important, as with every model of the ‘real world’, to give thought to the scope of the model (i.e. the system boundaries) and the level of detail (the subdivision/clustering of energy transforming steps). The system boundaries define what part of the ‘real world’ is captured in the block diagram. Therefore, the appropriate system boundaries need to be

14 | Page chosen depends on the purpose of the diagram to make a proper model of the input flows, output flows and intermediate conversions [2].

2) Operational profile of Thialf.

The real recorded operational profile of Thialf is provided by HMC, which includes two days’ onboard load consumption data during DP operation. A vast amount of details contained in the operational profile affords the author opportunity to calculate fuel consumptions, saving potentials and roughly estimate EES capacity for the new hybrid system. Answering this question is the most important foundation of the thesis.

3) EES configuration and Ragone plot.

Determining Thialf application requirements/targets for different technical specifications of EES device like energy capacity, power capacity and C-Rate is the basis of selecting right configuration and right technology for Thialf, which is one of the most important sub- questions. This question can be answered according to the outcome of sub-question 2) and meanwhile taking general EES (dis)charging characteristics into account.

Ragone plot is a log-log plot used for general performance comparison of various energy storage devices. On such a plot the values of energy density (in W·h/l) are plotted versus power density (in W/l). Thus, best technology and alternatives used for ship storage can be initially selected according to the plot. For example, if it is necessary to deliver energy in nanoseconds, capacitors are the only real choice. Similarly if the energy must be stored for weeks, batteries are the only choice. But for long term storage, one must compare battery storage to simply using fuel. Between the extremes, other technologies may be the most appropriate [1].

4) Modelling and cost analysis.

Different component models from TU Delft and other sources such as diesel engine A4 model from maritime department of TU Delft, battery model from SimPowerSystems (SPS) tool box and battery cycle life model from Valentin [23] are used in this part. System integration approach is used here to model the benchmark and hybrid power system. Thus based on the model, several control strategies of hybrid system can also be implemented to check its impact on system performance like fuel savings and EES cycle life. What’s more, diesel engine component (A4 model) can verify and improve the fuel consumption and saving results obtained from previous simple Excel model. Battery component (SPS model) can verify and improve the battery capacity and other specifications data got from previous analysis. Answering this question means the technical feasibility of the hybrid system can be determined.

Cost is always of concern, and sometimes is of overriding importance, even at the expense of performance (refer to 2.4). Thus cost analysis is one of the most important part of this thesis and is completed with support from HMC cost estimation engineers. After accomplishing all the above sub-questions, cost analysis can be carried out to analyze the investment cost, operational cost, maintenance and environmental influences

15 | Page of the new system. Comparing the cost and savings of proposed hybrid system leads to the answer of economic feasibility. When question four is answered, so is the main research question and as a result, it becomes possible to identify the feasibility of this envisage from both technical and cost related aspects.

2.5.2 Outline of the thesis In order to answer the research questions that were posed in the previous sub-chapter, the research that is performed follows the path that is described in flow chart Figure 2-5. Basically the thesis is designed to execute according to this flow chart by answering sub- assignments one by one and step by step.

Firstly, this thesis briefly summarizes previous work on research of how to use non-fuel alternatives of EES technologies in marine sector. Secondly, problem definition, main research objective and scope is finished in this chapter. Thirdly, selecting the best EES technology fulfilling Thialf on board requirements and developing tailor-made technical solutions on the way they are integrated into the system is accomplished according to operational profile of Thialf. Fourthly, a Simulink model of power system is developed to verify and improve the result of previous estimations. Last but not least, cost estimation is another important part in the thesis. When answering all these relative aspects a conclusion can be made to the research objective.

Additionally, an extra part of overviewing the functions and recent developments of EES technologies in different industries is initially explored in priority during the literature research phase since its unique importance. This process endows the writer a basic understanding of different EES technologies and also a better vision of this project, which is the starting point of this work. The content of this aspect is not included in the report as an independent chapter as it is not the main objective of this project. However, the knowledge I learnt during this process is used everywhere in this work. And there are a large number of papers, reports and thesis academically and commercially available on this specific subject thereby readers who are interested in this part can refer to other sources.

16 | Page

Figure 2-5 Research structure

2.6 Research scope

It is emphasized here that the research target of this work is the Thialf vessel; which is a vessel that has successfully been operated for 30 years. Thereby systems and components onboard are fairly reliable. Considering the main objective and time limit of this project the pre-set scope of this work by HMC can be summarized as:

1. No major modification of existing system;

2. No replacement of component installed onboard;

3. Feasibility study is the objective and first priority in this work instead of optimize the benchmark system and hybrid power system;

4. Replacing one DG (not two) in each engine room is the boundary of hybridization;

5. Commercially available and maturely developed EES technologies are of interest.

In general, discussions concerning EES implicitly refer to situations in which a storage technology other than fuel is added to the ship. But it is also very important to consider

17 | Page initially when the additional technology might be necessary or not. Because when it is not necessary the cost, size, and weight of the power system are all smaller [1]. For example, using new DE technology which have better low load performance can be a potential solution to the defined problem as well. This important method can be used especially during the design phase of a new crane vessel, however, it is not compatible with the research scope in this work.

What’s more, during the process of this project, Thialf on board systems are studied, understood and analyzed. Thus, engineers from equipment support group in HMC and the author are also motivated to look for more potential improvements of the benchmark power system which leads to other findings with respect to fuel savings like introducing variable frequency drives for sea water pumps, installing heat recovery system of utilizing exhaust gas thermal energy and etc. But the author is exclusively focused on the EES feasibility study. In addition, when one storage technology is being selected to install for good reasons on Thialf, that capability opens the door to using the technology for providing multiple services due to different control schemes over their lifetime. The possibility of other beneficial uses of the storage system besides the main functions of the EES system should be explored, which is not in the scope of this thesis. For example, EES units can also be used in a variety of ways like improve power quality8 on board.

Last but not least, working principles of EES system are supposed to be dealt within the thesis but it is not the emphasis to handle this into an advanced detail level. For example, a lithium-ion battery is dealt as the smallest entity for analysis, sub-division are not further made to reach details like electrode material, electrolyte and etc. Because in lithium-ion batteries, a large number of electrolytes and combinations of electrodes materials, which lead to different characteristics, can be used. With the large number of possible material combinations, there are still high development activities and until today it is not clear [24]. Anyhow, this is not the thesis of master students from chemistry or material science background. And for HMC, commercially available and maturely developed EES technologies are of interest in this project. Although the result of feasibility study to some extent is depend on developments of EES technologies as well, since it is a currently fast-growing and improving business [25].

2.7 Synthesis

Problem definition, research objective, research outline and research scope is completed

8 Power quality determines the fitness of electric power to consumer devices. Synchronization of the voltage frequency and phase allows electrical systems to function in their intended manner without significant loss of performance or life. The term is used to describe electric power that drives an electrical load and the load's ability to function properly. While "power quality" is a convenient term for many, it is the quality of the voltage—rather than power or electric current—that is actually described by the term. Power is simply the flow of energy and the current demanded by a load is largely uncontrollable.

18 | Page in this chapter. To sum up, “why”, “what” and “how” are all delivered in this chapter step by step, which is the map and compass of the author during the project. To emphasize at last, all the effort put into this work is to solve the main research objective:

A vessel is designed to perform a certain operational task: the mission. The mission of the vessel determines which functions are needed on board. Ship systems provide these functions. If necessary a system may be decomposed into sub-systems and ultimately into components [2]. With this approach, Thialf onboard systems can be decomposed and the most important system/components and connection between systems/components relative to the research objective can be found out, which offers a good basis for analyzing and modelling in following chapters.

3.1 Thialf functions

The starting point for functional analysis is the main goal or purpose of Thialf: the mission. A mission specifies which cargo to carry or which task to fulfil. Other specifications, such as the area of operation, speed and period of time at sea. Thialf’s two cranes provide for a depth reach lowering capability as well as a heavy lift capacity to install topsides. This multi-functional dynamic positioned DCV is customized for the installation of foundations, moorings, SPARs, TLPs, and integrated topsides, as well as pipelines and flowlines [20]. Thus, to sum up, important missions of Thialf are:

• Installation and removal of offshore and subsea structures.

• Pipelines and flowlines laying on the bottom of the sea.

Realization of the mission requires certain main functions including platform functions, operational functions, hotel functions and general support functions. The main functions group the underlying functions, which in turn may be divided into sub-functions as shown in function tree (incomplete) in Figure 3-1. The main function provide platform, for instance, can be divided into provide carrying platform, provide mobility and provide survivability. Hotel functions group all the hotel facilities which make life for the 736 crew on board. The main function provide general support groups the support for more than one function like electric power supply, ballast system, the hydraulic power supply, the

19 | Page

fuel supply, and cooling and heating. Many functions require the same support like electric power supply or cooling [2]. The operational functions of Thialf is attached in another separate function tree as shown in Figure 3-2.

Function tree visualizes the preceding functional breakdown of the machinery plant and to illustrate the energy flow on board. The operational functions and electric power supply is the main subject relevant to this thesis.

Thialf mission

Provide platform

Provide carrying platform

Provide buoyancy

Provide protected space

Provide mobility

Propulsion

Steering

Navigation

Provide survivability

Prevention of hazards

Detection of hazards

Provide hotel facilities Fighting of hazards

Provide accommodation

Provide food

Provide services

Helicopter deck

Provide general support

Electric power supply

Ballast system

Hydraulic power supply

Fuel supply

lubrication

Cooling

Control and monitoring

Compressed air supply

Provide operational functions

Figure 3-1 Thialf functional tree

20 | Page

Provide operational functions

Heavy lifting

Subsea construction

ROV

Positioning

Mooring and anchoring

Dynamic positioning

Positioning reference system

Pipe laying (optional)

Figure 3-2 Thialf function tree: operational functions

3.2 System & component

A system is the combination of machinery, equipment and its connections (piping, cabling) which performs the task indicated by a (sub-) function. Systems may be also grouped by their functions [2]:

• The propulsion system, the steering system and the navigation system provide mobility. Firefighting and detection systems and life-saving systems in accordance with the latest governmental requirement are examples of systems that provide survivability.

• Potable water systems, waste disposal systems, heating and air conditioning systems are the example of hotel systems. The helicopter deck which is suitable for a Boeing Chinook 234 is also included in the hotel systems.

• Support systems like electric power supply system, ballast system and fuel systems provide support functions.

• Operational systems provide the operational functions, e.g.: crane system, DP system and pipe laying system.

Systems then can be divided into sub-systems and components depending on the complexity and divisibility of the building blocks and the intended purpose of description and modelling. A component is a building block of a system. The diesel engine is an example of a component of the propulsion system to provide mechanical power, but in some cases it may be not sufficient to look at the diesel engine as a component like to model the heat transfer process on single cylinder can be split up in to different components. The goal of this chapter is not to illustrate the components’ function, its behavior and operational characteristics but its interaction between different components and energy flow in the system. Because it will be the basis to integrate EES systems on board in a well-founded manner both technically and economically.

Thialf is equipped with electric propulsion, a combined electric power plant provides

21 | Page power to the propulsion motors and other electric systems. Thus in this sub-chapter main components of electric power supply system are dealt with as they are the most important objects of the thesis. On board Thialf the electric power plant converts chemical energy in fossil fuels into electric energy. In general, this is a three-stage process in which the following conversions can be distinguished:

• Energy contained in fossil fuels into mechanical energy (diesel engine),

• Mechanical energy into main electric energy (generator),

• Main electric energy into secondary electric energy (distribution and conversion).

Therefore, the main components that will be dealt with in the remainder of this sub- chapter: diesel engine, generator, distribution and conversion and users as illustrated in Figure 3-3.

Distribution Diesel engine Generator & Users Conversion

Figure 3-3 Main components of an electric power plant (Source: [2])

The single line diagram of the HV (High voltage) power distribution is schematically shown in Figure 3-4. It can be seen that the basic installation of Thialf consists of 12 diesel generator sets which are housed in three separated engine rooms. Separated switchboard sections can be coupled with bus-ties. Main electric power supply to large consumers like thrusters and cranes as illustrated in the figure use a high voltage of 4160 V and 60 Hz. Other systems using 440 V 60 Hz LV secondary electric power distribution like diesel engine heating system are not shown in the figure.

G7 G5 G3 G1 G2 G4 G6 G8

MSB1-PS MSB1-SB

CR CR G9 G10 G11 G12

MSB2-PS MSB2-SB MSB3

S1 P2 P3 S2 S3 P1

Figure 3-4 HV single line diagram of power system

3.2.1 Diesel engine Table 3-1 gives the engine information. Note that Engine 1-6 were installed on Thialf when she was delivered in 1985 and Engine 7-12 were installed years later. What’s more

22 | Page

Engine 7-12 are initially produced for propulsion, thus at rated 510 rpm, each cylinder produce 750 kW rated power for propulsion. But now all the engines runs at 514 rpm to rotate the rotor of generator to produce 60 Hz alternating current.

Table 3-1 Diesel engine information Information Engine 1-6 Engine 7-8 Engine 9-12 Type MAK 8M552 Sulzer 8ZAL40S Sulzer 6ZAL40S Manufacturer Ube Industries Ltd Wartsila Wartsila Rated speed 514 rpm 510 rpm 510 rpm Rated power 4898 kW 6000 kW 4500 kW Combustion cycle 4-stroke cycle 4-stroke cycle 4-stroke cycle Number of cylinders 8 8 6 Fuel type on board MDO MDO MDO Year 1985 1998 (installed on - (installed on Thialf in 2010) Thialf in 2007)

3.2.2 Generator There are in total 12 synchronous generators driven by diesel engines which installed in the 3 engine room. Generators information are shown in Table 3-2.

Table 3-2 Generator information Information Generator 1-6 Generator 7-8 Generator 9-12 Phase 3 3 3 Frequency 60 Hz 60 Hz 60 Hz Speed 514 rpm 514.3 rpm 514.3 rpm Power 6570 kVA 7269 kVA 5429 kVA Output 4600 kW 5815 kW 4343 kW Power factor 0.7 0.8 0.8 Voltage 4160 V 4160 V 4160 V Nominal current 912 A 1009 A 753 A Poles 14 14 14

3.2.3 Distribution & conversion

Main switchboards receive, control and distribute electric energy from generators to loads and secondary electric energy supplies [2]. AC-DC-AC converters that convert three phase AC with 60 Hz constant frequency and 4160 V constant voltage to a three phase with varying frequency and voltage, which in order to power and control the speed of motors for thrusters, are not shown in Figure 3-4. And also the AC-DC converters for powering and controlling the speed of DC motors for the cranes are not shown. These converters after the main switchboard can act as a buffer against electric distortion in the net as well. In addition, during different operation modes the main switchboard can be connected and separated with each other according to different requirements of redundancy, Table 3-3 shows the number of high voltage (4160 V) grids and running diesel generators in different modes. It should be noted that in DP and heavy lifting

23 | Page operation it is assumed that three DGs are running in AFT engine room but actually there are two DGs running according to two days’ recorded data on 20th and 24th June 2015.

Table 3-3 Estimation of power configurations in different operation mode NO. HV-grids Standby Transit Standby DP-2/3 Heavy lifting inshore DP-1 Operation on DP-2/3 Grid 1 PS+SB+Aft PS+Aft(PS) PS PS PS Grid 2 - SB+Aft(SB) SB SB SB Grid 3 - - Aft Aft Aft Typical NO. of 3 10 9 9 11 running DG Grid 1 3 5 3 3 4 Grid 2 0 5 3 3 4 Grid 3 0 0 3 3 3

3.2.4 Users As grouped in chapter 2.1, consumed loads can be divided into hotel load, propulsion load and crane load. A detailed introduction is given below.9

Propulsion motors The induction machine is the most rugged and the most widely used machine in industry [26]. Six induction/asynchronous motors are used to drive the propellers of the thrusters. The rotation speed of the thruster motors is controlled by frequency converters and frequency converters are powered by two parallel transformers. Table 3-4 shows the information of motors (converters and transformers are not shown).

Table 3-4 Thruster driver information Information Motor 1-6 Phase 3 Power 5500 kW Power factor 0.84 Speed 0-520 rpm

Crane motors DC motors are used to drive the crane. There are 8 swing motors, 4 hoist motors and 2 boom motors for each crane. Its speed can be controlled over a wide range with relative ease [26].

9 As the constant and mix nature of hotel load, its introduction part is neglected.

24 | Page

Table 3-5 Crane motors information Information Main hoist & boom hoist Swing motors motors Rated output 1250 HP (932 kW) 525 HP (392 kW) Voltage 600 V 600 V Current 1680 A 708 A Speed 400/1000 rpm 500/1000 rpm

3.3 Energy flow diagram (EFD)

To achieve overall efficiency, an integrated approach is required. If each component only delivers its own bit in isolation, it is like separate bricks. Thus, it is more important to bring all the separate bricks together to build a complete effective energy system. Therefore, after the functional decomposition of Thialf, the energy transfers and conversions from energy source (ES) to energy use (EU) can be modelled in simple energy flow diagram. What’s more, as with every model of the “real world”, the scope of the model (system boundaries) and the level of detail (degree of subdivision or clustering of energy transforming steps) are also considered in this diagram. This provides the vital basis for modelling in Simulink in proceeding chapters.

3.3.1 Types of energy and conversion Table 3-6 below briefly introduces types of energy and types of energy conversion in a general energy flow diagram. Like fuel (MDO) is kind of chemical energy source. In this thesis X/Y is the abbreviated notation of an energy converter: X represents the incoming energy flow and Y represents the outgoing energy flow.

Table 3-6 Matrix of possible energy conversions (Source: [2]) To → Energy Mechanical Electric Hydraulic Pneumatic Heat (Q) From ↓ source (ES) energy (M) energy (E) energy (H) energy (A) Free piston ES Reformer Engine Fuel cell Burner unit Gearbox & Hydraulic Air Chiller M E-Generator Propeller pump compressor unit Electric E Electrolysis Converter Heater motor Hydraulic Hydraulic H motor transformer Pneumatic A Reducer motor Heat Q Turbine exchanger

The conversions at off-diagonal locations in the matrix are conversions from one type of energy to another type of energy. The main diagonal, however, contains conversions that

25 | Page do not change the energy type but changes its characteristics, such as voltage, frequency or angular velocity. An example shown in the matrix is the electric converter that can have the following functions: (1) convert AC electric energy to DC electric energy and vice versa, (2) change the frequency of AC electric energy and (3) change the voltage of electric energy [2]. After introducing types of energy and conversion the next step is to build the EFD on board Thialf.

3.3.2 EFD on Thialf Thialf uses integrated power supply on board: thrusters and its auxiliaries, and other big consumers of electric energy draw from a common supply. Figure 3-5 shows the EFD on Thialf. It should be noted that in different operation modes EFD can be different. The EFD below models the standby inshore operation mode which PS, SB and AFT switchboard are connected. The part within red lines of dashes is of core interests and importance to this thesis.

Main Main High voltage Elec. Main Gear ES eng. gen. AC bus Propeller Propulsion 1-12 1-12 4160V 60 Hz Contr. motor -box

ES/M M/E E/E E/M M/M M/M M

Elec. Crane Gear Mechanical ES/M M/E Contr. motor -box consumers

ES/M M/E E/E E/M M/M M

Elec. Large consumers Mechanical ES/M M/E Contr. like ballast pump consumers

ES/M M/E E/E E/M M Legend ES/M M/E 400 V AC bus X Storage of Elec. Elec. Mechanical energy X ES/M M/E Transformer Contr. motor consumers Conversion from X/Y energy X to ES/M M/E E/E E/E E/M M energy Y Electrical Ultimate use of ES/M M/E Converter X consumers energy X

ES/M M/E E/E E Transport of Electrical energy ES/M M/E Converter consumers

ES/M M/E UPS E/E E Distribution

Secondary bus

Figure 3-5 EFD of Thialf power system

The input flow, one of the system boundaries is chosen to be the ES because the chain of conversions on board starts with the conversion of chemical energy stored in MDO. The output flows are the types of EU. The power for thruster motors and crane motors are the most important power use. Therefore, the power supplied to different motors, delivered output power is the appropriate system boundary. But it should be mentioned that only the total consumed load can be recorded on Thialf, thus, output power of the generator (real power) is the system boundary instead of different components’ load.

26 | Page

The system boundaries define which part on board of Thialf will be modelled. After that, the degree of subdivision of energy transformations in the diagram needs to be considered. The above EFD presents the diesel engine as a single energy transformation from chemical energy into mechanical energy: it would be an appropriate level of detail since the purpose of this thesis is to illustrate the overall function of the plant. It needless to be presented in a higher level of detail as: E/Q/M, which shows different energy transformation steps of the plant, i.e. combustion, heat transfer and power generation in the cylinder [2].

3.4 Synthesis

After modelling the energy transfer and distribution of Thialf in EFD, an initial ideal of integrating EES technologies into onboard power system can be illustrated in in Figure 3-6 and the EFD of hybrid power system is shown in Figure 3-7. A main EES system can be connected to the main switch board to realize its function as mentioned before. Additionally, another extra EES device might be installed on board to collect regenerative energy from lowering operations of cranes, which makes no influence on the charging/discharging strategy of the main EES system. But obviously, the regenerative energy can also be directly stored by the main EES system. However, this configuration may introduce problems because the daily operation of the crane is irregular, which means when the main EES system is discharging it may cannot be used for charging (storing energy from regenerative process) at the same time. What’s more, how to use the regenerative energy depends on the potentials of storing energy based on cranes’ daily operation of Thialf, which is dealt with in the preceding chapter. This EFD of the hybrid system is an initial design, which is necessary to be modified according to the details of thialf power system, operational profile and current EES technologies.

G7 G5 G3 G1 G2 G4 G6 G8

MSB1-PS MSB1-SB

E/E E/E

CR CR EES G9 G10 G11 G12 EES

MSB2-PS MSB2-SB MSB3

E/E S1 P2 P3 S2 S3 P1 EES

Figure 3-6 HV single line diagram of hybrid power system

27 | Page

Main Main High voltage eng. gen. AC bus Elec. Main Gear ES Propeller Propulsion 1-12 1-12 4160V 60 Hz Contr. motor -box

ES/M M/E E/E E/M M/M M/M M

Elec. Crane Gear Mechanical ES/M M/E Contr. motor -box consumers

ES/M M/E E/E E/M M/M M

Elec. Large consumers Mechanical ES/M M/E Contr. like ballast pump consumers

ES/M M/E E/E E/M M

400 V AC ES/M M/E Converter E/E bus Elec. Elec. Mechanical ES/M M/E Transformer Contr. motor consumers

ES/M M/E E/E E/E E/M M

Electrical ES/M M/E Converter EES consumers

ES/M M/E Transformer Converter E/E E

Electrical ES/M M/E E/E E/E Converter EES consumers

ES/M M/E UPS E/E E

Secondary bus

Figure 3-7 EFD of Thialf hybrid power system 4. Operational profile

The operational profile of Thialf means an oversight where the size and duration of power demand is known under different kinds operational modes, which is of vital significance to carve out what’s needed from traditional DG sets and another kind of EES technology in the hybrid system – whether that’s a e.g. battery pack, flywheel or supercapacitor etc. With considerations of what (power) is expected from EES device at every operational point, the initial energy capacity of EES device can be calculated. Initial capacity can be improved considering state of charge (SOC) and EES energy efficiency in the next chapter. Given capacity in combination with (dis)charge power, other specifications of EES like daily full equivalent cycles can be estimated, which provides the basic technical criteria on selecting the most promising technology for Thialf. Last but not least, fuel saving potentials of hybrid power system is also estimated by assuming an ideal EES system.

4.1 General

It was explained in 2.1 that there are four operational modes for Thialf (see Figure 2-2):

28 | Page

Standby Inshore, Transit, Standby DP-1, and DP-2/3 Operation. Since in terms of standby inshore and transit mode, there is no redundancy requirement and load consumptions are always very steady. Therefore, the fuel saving potentials by using EES in these two modes are eliminated. In other words, EES (or hybrid system) is exclusively designed for DP mode as DP and heavy lifting operation is the root reason of low load with variations. Additionally, mode characteristics are summarized in Table 4-1. As a result, two days’ data of Thialf (20th and 24th, June, 2015) – second by second recorded load consumption in DP mode (including heavy lifting operations) is provided by HMC, which offers the vast wealth of details for analysis e.g. derive design criteria for EES in this chapter.

Table 4-1 Mode characteristics

Mode Redundancy DE load Potentials for fuel requirement saving using EES Standby inshore No Medium & steady No Transit No High & steady No Standby DP-1 Yes Low & spiky Yes DP-2/3 Yes Low & spiky Yes

It should be also noted that it is assumed in this work that the recorded load consumption profile in DP mode is a “typical” DP load profile. “Typical” means it is representative of DP mode in terms of environmental conditions, thrusters and crane operation. Definitely, different factors like environmental conditions can always change which will influence the onboard load consumption. Therefore, more days’ data e.g. one month are needed to verify or improve results which derived from these two days. But in this project, more recorded data is not available. Thus, the verification and improvement work on this point is arrange to recommendations for future investigation.

4.1.1 DP mode In this sub-chapter, details of load consumption characteristics during DP and heavy lifting operation are described carefully.

PS & SB load consumption 15000 PS SB DG1 capacity DG1 and 3 capacity DG1,3 and 7 capacity 13000

11000

9000

7000 Load(kW)

5000

3000

1000 0 0.72 1.44 2.16 2.88 3.6 4.32 5.04 5.76 4 Time(second) 10

Figure 4-1 PS & SB load consumption day1

Figure 4-1 shows the recorded day-1 generator load consumption in PS and SB engine

29 | Page room. As illustrated in the Figure 4-1 and mentioned in chapter 2.1.5, there are three DGs running in each engine room during this mode to secure the required power for heavy lifting and fulfill strict redundancy requirements. According to the recorded data, there are seldom occasions that consumption load exceeds one DG’s capacity although given that the weather condition during the recorded hours might be quite good. This extremely conservative power strategy of Thialf implies the extremely low load of diesel engines. What’s more, it can be noticed that the load variations due to heavy lifting and thrusters are still prominent in this nice weather condition, which is the same for AFT engine room as illustrated in Figure 4-2. It should also be noted that load variations in AFT engine room are purely induced by thrusters.

AFT load consumption

AFT DG9 capacity DG9 and 12 capacity

7000

5000 Load(kW)

3000

1000 0 0.72 1.44 2.16 2.88 3.6 4.32 5.04 5.76 4 Time(second) 10

Figure 4-2 AFT load consumption day1

Statistical data of both DE’s operational (load) profile and ramp rate can be derived from above recorded data, which quantify the low load condition and fluctuations of each engine as well. Results are shown in Figure 4-3 and Figure 4-4. Based on Figure 4-3, it is deduced that the ship main DE usually operates at low loads around or lower than 20%, and as a result, very far from the optimal operational point, which indeed indicates bad fuel economy. In terms of quantifying dynamics of DE transient load, ramp rate is been used in this work. General a DE is able to cope with a load increase from 0% to 100% in 20-25 minutes. Converting the general performance to ramp rate for Thiaf DEs, it is between 200 kW/s to 250 kW/s. Higher ramp rates are also possible for short intervals e.g. 500 kW/s for a load increase from 30% to 50% engine load (not the full range from 0% to 100%). Figure 4-4 shows that during more than 99% of recorded duration ramp rate value is less than 80 kW/s although the maximum recorded value can achieve 500 kW/s (lasting for a second) which is about 10% of DE’s nominal capacity. Therefore, transient engine load variations are fairly “acceptable”10 except demanding operation periods of heavy lifting. But it should be noticed that 80 kW/s is still higher than the ideal

10 However, it is not clear what the impact of such load variations are on fuel consumptions, which will be tackled in chapter 6.

30 | Page rate11 for ideal DE operations.

DE operational profile (two days data) 60

10 Percentage of working time [%]

0 0 10 20 30 40 50 60 Load[%MCR] Figure 4-3 Diesel engine operational profile

5 10 DE Ramp rate(two days data) 2

0 0 20 40 60 80 100 120 140 160 180 200

20 Cumulative time[s]

0 200 240 280 320 360 400 440 480 520 Ramp rate[kW/s]

Figure 4-4 Diesel engine ramp rate

4.1.2 Heavy lifting operation Figure 4-5 showing the consumed load during topside installation process is a zoomed in part of Figure 4-1 and Figure 4-2. Studying this figure helps to understand detailed crane operation process during cargo lifting and lowering. There is a load ramping up of nearly 5 MW in the PS and SB engine room from 30500 second to 31150 second, which indicates a lifting process of topside from barge by cranes of Thialf. From then on to around 40000 second, cranes were using brakes to hold the topside, meanwhile, Thialf was being maneuvering on DP towards the jacket. During this process, there is no DC current flowing in the motors of the crane, thus, the load keeps at a relative low level as shown in the figure. But there are still apparent load variations because of varying forces from surrounding environment. Then, motors of crane came in to take over the load from brakes and also hoist the topside to the correct installation height, which is the reason for load increase from around 40600 second to 40800 second in the PS and SB engine room. Finally, it was the set-down operation, which the crane loaded the topside to the jacket gradually from 40800 second and last for around 15 minutes. In addition, two small peak loads each sustained for about 10 minutes between the lifting and lowering

11 One of the perfect operation conditions for diesel engine is running under constant load, which is impossible in most situations. Therefore, an ideal ramp rate margin of 10 kW/s is set for each engine in the new hybrid system with EES.

31 | Page operation is possible for extra crane operations like slightly adjusting the holding height of the topside.

Load consumption 8000 PS SB AFT

7000

6000

5000

4000 Load(kW)

3000

2000

1000 3.05 3.23 3.41 3.59 3.77 3.95 4.13 4 Time(second) 10 Figure 4-5 Load consumption during heavy lifting

4.2 Load data analysis

A series of detailed computations are implemented in this sub-chapter according to replace one DG with EES system in each engine room. Different technical configurations of EES can be derived from different “design reference” (see Table 4-2) based on different EES functions e.g. the EES capacity can be measured in two dimensions, power capacity and energy capacity, respectively.

There are three defined EES functions according to 2.3: spinning reserve, enhanced dynamic performance and store regenerated energy. The enhanced dynamic performance function is sub-classified into typical peak shaving function and demanding peak shaving function, which outlined in Table 4-2. Typical peak shaving function corresponds to the typical DP mode defined in 4.1, which means the configuration of EES for this function is derived from the recorded two days’ typical load consumption profile. The demanding peak shaving function is defined according to demanding crane operations e.g. the crane load ramp up to its nominal capacity of 5.5MW rapidly (refer to 4.2.5).

Table 4-2 EES function and configuration reference EES Function Design reference Spinning reserve Assumed DE failure scenario in DP mode Typical peak shaving Recorded “typical” load consumption in DP mode Demanding peak shaving Assumed demanding crane operational profile Store regenerated energy Assumed load-lowering scenarios 4.2.1 Spinning reserve The predefined period of time for spinning reserve function is (at most) 5 minutes as mentioned in chapter 2.3. A scenario analysis on DE failure is implemented below. Normally, DGs are sharing load and when the load is higher than 80% of its capacity, power management system (PMS) will bring another DG online. Thereby, in the new

32 | Page hybrid power system, assuming load of PS engine room is 7360 kW (2*4600 kW*0.8), which is exactly provided by two DGs (NO.1 and NO.3) running at 80% capacity as shown in the Figure 4-6. At 300th second, NO.3 DG fails (as illustrated by green dotted line in Figure 4-6), after which load of NO.1 DG increases to 100% gently (10kW/s, red solid line in Figure 4-6) and starting process of NO.5 DG is immediately activated by PMS (as illustrated by cyan line with cycles in Figure 4-6). Orange dotted line in Figure 4-6 stands for the power delivered by EES device. According to the figure, EES devices certainly begin to discharge immediately after NO.3 engine fails. What’s more, NO.1 and NO.5 engine are controlled to run both at 100% for a certain period in order to charge the EES. Thereby the minimum energy capacity and average/maximum power rating of EES can be calculated for this scenario according to EES load (orange dotted line) in PS engine room. The same scenario analysis can be implemented also in SB and AFT engine room.

Additionally, it should be noted that, the assumed PS engine room load and 5 minutes starting time of standby DG is conservative for determining the EES size in typical DP mode. Because according to Figure 4-1 and Figure 4-2, assumed load is much higher than recorded typical load and standby DG can be brought online within shorter period e.g. 2 minutes.

Last but not least, in the hybrid power system of AFT engine room, there is only one DG running. Thereby if it fails, the load of AFT engine room is entirely handled by EES device. For this reason, the minimum energy capacity of EES in AFT is 273 kWh, which is larger than that in PS and SB engine room. But if the standby DG is set to be brought online in 3 minutes after failure of the running DG, which can be easily implemented by PMS, then the minimum energy capacity of EES can be reduced to 230 kWh. Results is shown in Figure 4-6 and Table 4-3.

EES provide redundancy (PS) E.R. load 8000 NO.1 DG load NO.3 DG load Start process of NO.5 DG 6000 NO.5 DG load EES load

4000

2000 Load(kW)

-2000 0 120 240 360 480 600 720 840 960 1080 1200 Time(second)

Figure 4-6 EES spinning reserve function for PS or SB

33 | Page

Table 4-3 Design margin for spinning reserve Specification PS/SB AFT Minimum energy capacity 227 kWh 230 kWh Peak power 3680 kW 3440 kW Average power 2789 kW 2756 kW

4.2.2 Typical peak shaving - energy capacity Energy capacity (in kWh) of EES device is the most important technical parameter need to be determined. Analysis steps and calculation method is discussed with assistance of Figure 4-7. Firstly, a 15 minutes’ load consumption map in AFT engine room without EES is plotted using gray polyline in the figure. Secondly, one DG in AFT engine room is replaced by EES device, thus there is only one DG running with assistance of EES device in the new hybrid system. What’s more, a simple control algorithm for DG is implemented to keep its ramp rates within 10 kW/s, which is explained in the end of 4.2.1. This way, the load of DG in the hybrid system can be determined and plotted in red polyline as illustrated in the figure. Thirdly, power difference between grey and red line in the figure, therefore, indicates extra peak power shaved by EES device through discharging or surplus power used to charge the EES device, which is shown by green line in the figure. Green line is the EES discharge/charge line which means when it is higher than 0 EES device is discharging, otherwise when it is lower than 0 EES device is been charged.

AFT load map with EES discharge & charge 3000 AFT load DG load EES discharge/charge y=0 2500

2000

1500

1000

Load(kW) 500

-500

-1000 3.69 3.708 3.726 3.744 3.762 3.78 4 Time(second) 10 Figure 4-7 15 minutes’ AFT load map in hybrid system (based on day1)

Using this method, EES (dis)charge power in AFT on complete timeline of day1 can be plotted as shown by blue polyline in the figure below. Integrating the (dis)charge power of EES leads to the cumulative discharge (positive) and charge (negative) energy as shown by red line in Figure 4-8.

34 | Page

EES discharge/charge in AFT 2500 25 EES discharge/charge power EES discharge/charge energy

2000 20

1500 15

1000 10

500 5 Power(kW) Energy(kWh) 0 0

-500 -5

-1000 -10 0 0.72 1.44 2.16 2.88 3.6 4.32 5.04 5.76 4 Time(second) 10 Figure 4-8 EES discharge/charge level in AFT (based on day1)

In the last step, assuming EES has the largest cumulative discharge energy (20 kWh as shown in Figure 4-8) as its initial energy level. Subtracting discharge and charge energy from initial level results in the second-by-second energy level within EES device as plotted in red polyline in Figure 4-9, of which the maximum value indicates the minimum energy capacity of EES in AFT engine room without considering SOC and roundtrip efficiency.

EES energy capacity in AFT 2500 35 EES discharge/charge power EES cumulative capacity

2000 30

1500 25

1000 20

500 15 Power(kW) Energy(kWh) 0 10

-500 5

-1000 0 0 0.72 1.44 2.16 2.88 3.6 4.32 5.04 5.76 4 Time(second) 10 Figure 4-9 EES energy capacity in AFT (based on day1)

Control algorithm An initial control algorithm is invented to obtain the minimum EES energy capacity. Since one of the main designed EES function is to absorb load variations to protect DE from seeing the fluctuations directly. Thus at first, a ramp rate interval such as from -7kW/s to +7kW/s is set for DE, which means transient load variations for DE is less than 0.2% of its nominal capacity. Using the same method as before, minimum EES capacity can be calculated as illustrated in Figure 4-10. It can be seen that EES energy level continues descending to 0 rather than keeping stable as in Figure 4-9, which indicates a minimum capacity of 209kWh. Moreover, given this is still not a complete time frame for one day, larger capacity should be expected.

35 | Page

EES energy capacity in AFT 2500 EES (dis)charge power 200 2000 EES (dis)charge energy EES energy level

1500 160

1000 120

500 80 Power (kW) Energy (kWh) 0

40 -500

-1000 0 0 0.72 1.44 2.16 2.88 3.6 4.32 5.04 5.76 4 Time(second) 10 Figure 4-10 EES energy capacity in AFT (hybrid mode 1)

Thus, a new control algorithm is developed based on this one in order to reduce the EES energy capacity. The root reason of minimum required EES energy capacity becoming too large finally according to Figure 4-10 (red line) is that EES cumulative (dis)charge energy continues to increase with time. Thereby, it is envisaged that keeping the cumulative (dis)charge energy (or energy level) more or less constant (small variations can be allowed) may solve this problem. The new control algorithm is designed according to this idea. Firstly, the deviation from pre-set constant energy level of EES is divided into three categories which outlined in Table 4-4.

Table 4-4 EES energy level EES cumulative Deviation status (dis)charge energy >1 kWh Positive (+) -1 to 1 kWh Neutral (0) <-1 kWh Negative (-)

The control algorithm is explained using a simplified schematic diagram which is Figure 4-11. The blue solid line represents the power of two DEs in AFT engine room in benchmark mode. And the red solid line is the power of the single DE with medium deviation level, which shows an absolute ramp rate of 7 kW/s. What’s more, as shown in the figure the absolute ramp rate of DE can be set higher to 10 kW/s or lower to 4 kW/s according to positive/negative deviation status. In this way, especially during big load spikes and valleys DE can ramp up and down steeper or steadier but within the limit of 10 kW/s so that the EES energy level is more stable and thereby the minimum required energy capacity is much smaller as shown in Figure 4-9.

36 | Page

AFT control algorithm (example) DE power benchmark 600 DE power hybrid @ 0 400 deviation status

200 DE power hybrid @ + deviation status

0 DE power hybrid @ - 0 20 40 60 80 100 120 deviation status

Power [kW] -200 DE power hybrid @ - -400 deviation status DE power hybrid @ + -600 deviation status Time [Second] Figure 4-11 EES control algorithms

Complete results of minimum energy capacity of EES for each engine room based on two days’ data is shown in Table 4-5 and also figures in Appendix A. It can be concluded from Table 4-5 that 78 kWh is the design value of available energy which can be delivered by PS and SB EES device (without considering EES SOC and energy efficiency). In terms of AFT engine room, since no crane is connected thereby this value is 47 kWh which is lower than that for PS and SB engine room.

Table 4-5 Minimum EES energy capacity Engine room Day1 Day2 AFT 33 kWh 47 kWh PS 56 kWh 78 kWh SB 50 kWh 22 kWh

4.2.3 Typical peak shaving - power Power, which describes the rate of electric energy transferred per unit of time by an electric circuit to/from the EES system, is frequently a factor when determining how many or what kind of EES cells are required for a given application. For example, one may want to know whether a particular battery can power a cell phone, or how many cells are required to make a pack that can power a car to provide the desired acceleration [27]. Statistical data on the magnitude of discharge power and charge power of EES device can be derived from EES discharge/charge line and as shown in Table 4-6.

37 | Page

Table 4-6 EES discharge & charge power EES discharge Counts Counts Counts Counts Counts Counts power (kW) PS day1 SB day1 PS day2 SB day2 AFT day1 AFT day2 0 to 100 18069 14499 17122 15296 20393 24585 100 to 200 1069 1142 1549 1400 832 1630 200 to 500 1377 1383 1671 1439 998 1077 >500 574 581 992 725 758 1035 Avg. power 63 kW 73 kW 100 kW 86 kW 66 kW 79 kW Max. power 2282 kW 2273 kW 3079 kW 1967 kW 2494 kW 2315 kW EES charge Counts Counts Counts Counts Counts Counts power (kW) PS day1 SB day1 PS day2 SB day2 AFT day1 AFT day2 0 to 100 19583 25017 25368 30822 23137 26029 100 to 200 1190 1280 1957 2155 1442 2352 200 to 500 1336 1422 1551 1608 1961 2306 >500 500 368 772 331 321 694 Avg. power 59 kW 46 kW 71 kW 47 kW 56 kW 71 kW Max. power 2253 kW 2206 kW 2547 kW 1005 kW 906 kW 1251 kW

Average power is calculated using Equation 4-1:

( ) ( ) = Equation 4-1 𝑑𝑑𝑑𝑑𝑑𝑑( 𝑐𝑐ℎ)𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡 𝑡𝑡 ∫0 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 Thus, it can be concluded that although𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 most of times (dis)charge power is quite low for 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 𝑑𝑑𝑑𝑑𝑑𝑑 𝑐𝑐ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 𝑑𝑑𝑑𝑑𝑑𝑑 𝑐𝑐ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 EES device, however, the maximum (dis)charge power can be quite high. The current handling capability of EES device thereby should be designed so that EES can successfully delivery such high power for the required period.

4.2.4 Typical peak shaving - cycle life In many applications the EES device is expected to maintain its major properties over many discharge – charge cycles. Cycle life – the number of cycles which can be delivered by an EES system under specified conditions before its performance fails to meet specified criteria, e.g. its capacity has become too degraded of typically 70-80% of initial capacity for batteries. Due to nonlinear physical effects in the battery, this depends upon a number of factors, especially temperature and usage pattern like depth of discharge (DOD) and (dis)charge rate in each cycle [28]. For this reason, the cycle life specification can only be used as guideline as real-life operating conditions rarely match those used in deriving the cycle-life specification [29]. Thus, it is critical to calculate cumulative energy throughput of EES installed on Thialf, so that the full equivalent cycles per day can be obtained using Equation 4-2 after energy capacity is settled.

= Equation 4-2

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑡𝑡ℎ𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟ℎ𝑝𝑝𝑝𝑝𝑝𝑝 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 Results𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝐸𝐸of𝐸𝐸 𝐸𝐸calculation𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 is𝑜𝑜𝑜𝑜 shown𝐸𝐸𝐸𝐸𝐸𝐸 in Table 4-7. 𝐸𝐸𝐸𝐸𝐸𝐸 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐

38 | Page

Table 4-7 Cumulative energy throughput & full cycles Items PS day1 SB day1 PS day2 SB day2 AFT day1 AFT day2

Cumulative 5.8 hours 4.9 hours 6 hours 5.3 hours 6.3 hours 7.9 hours discharge time

Cumulative 6.2 hours 7.7 hours 8.3 hours 9.7 hours 7.4 hours 8.8 hours charge time

Recorded time 16.9 16.9 19.6 19.6 16.9 19.6 hours hours hours hours hours hours

Discharge/charge 71% 75% 73% 77% 81% 85% portion

Cumulative 365 kWh 354 kWh 584 kWh 449 kWh 418 kWh 616 kWh discharge energy

Cumulative 365 kWh 354 kWh 584 kWh 449 kWh 418 kWh 616 kWh charge energy

It can be concluded from the table that EES device in each engine room is required to continuously (dis)charge which occupies around 80% of the recorded duration. Therefore, in terms of EES device, it is designed in a more conservative way that average (dis)charge power is 100 kW (maximum value according to Table 4-6 and it is set to a continuously (dis)charge profile for 24 hours per day which means (dis)charge duration is 12 hours for a complete typical DP-2/3 operation day. In this way, the estimated cumulative energy throughput is 1200 kWh per day.

4.2.5 Demanding peak shaving Although, a series of detailed and accurate calculation is implemented above based on recorded data in terms of typical peak shaving function. However, power margin of each crane connected to PS and SB engine room is 5.5 MW and load ramp rate margin is set to be 1 MW/s. Thus the idea is illustrated in Figure 4-12. Without EES device, three DGs have to be ramped up and down stressfully in 1 MW/s, however, two DGs can be ramped up and down gently in 20 kW/s with EES device. Thus extra energy delivered by EES or extra energy from DGs used to charge the EES device can be calculated and is shown by yellow and green color respectively in the figure. Result is outlined in Table 4-8. It can be noticed that minimum energy capacity and power rating of EES is largely increased due to crane operations at full load.

39 | Page

EES Take step load 9000 EES discharing

8000 charging EES system without EES

7000 system with EES

6000 Load(kW) 5000

4000

3000 0 120 240 360 480 600 720 840 960 1080 1200 Time(second)

Figure 4-12 EES enhanced dynamic performance function

Table 4-8 Design margin for enhanced dynamic performance Specification Value Minimum energy capacity 206 kWh Peak power 5.38 MW Average power 2.75 MW

4.2.6 Regenerated energy Since Thialf is fitted out with two 7100 ton capacity cranes, so the possible regenerated energy available for example from lowering a 1000 ton subsea module 800 m is potentially vast. In this case, control algorithms can be afforded so that regenerated energy is designed to feedback to the grid and other consumers like hotel loads which will effectively be acting as the crane’s braking system. But considering the extremely low frequency of such subsea installation operations and the unclear technical issues and investments to realize such a modification to the existing AC grid, this idea might not be so attractive.

More often, regenerative energy can be collected to charge EES device from frequent daily light crane operations like lowering a 20 feet container12 10 meters onto the working deck and deceleration of rotations13 instead of burning it off using resistors. Additionally, energy can be regenerated from lowering the boom of crane is neglected since the angular velocity is too small. The result of potential savings is shown in Table 4-9. Clearly, it can be seen that only a fairly small amount of energy can be regenerated from such crane operations. Thus, the initial designed function of using EES to receive the regenerated energy from cranes can be eliminated to reduce the complexity of the system.

12 Assuming the typical mass of a 20 feet container is 14 tons and period of time during lowering process is 2.5s. 13 Moment of inertia of Thialf main crane is 4299194000 kg·m2, the max swing speed is 0.5 rpm and the deceleration ramp is 4 seconds in emergency and 8 seconds in normal conditions.

40 | Page

Table 4-9 Design margin for regenerated energy Specification Lowering Rotation Boom Regenerative energy 0.38 kWh 1.64 kWh ≈0 Average power 550 kW 1473 kW ≈0

4.3 Fuel saving potentials

An initial estimation of fuel saving potentials for hybrid power system can be estimated before selecting alternative EES configurations according to SFC. SFC curves are fitted according to data obtained during factory trials on testing bed when DE was newly delivered as shown in Figure 4-13. It should be also mentioned that since DE engine in this project mainly running at around 15% low load conditions, thereby testing data at corresponding low load points like 25% and 10% are very crucial for a correct estimation. Additionally, assumptions are made in order to simplify the calculation: 1. SFC testing data used here are provided by engine manufactures which implemented tests on test bed inside the factory when the engine is delivered to the ship. 2. Influence of DE’s running time on its SFC is neglected. 3. Assuming that SFC during transient dynamic load variations can be estimated using steady-state values shown in Figure 4-13. 4. Efficiency of EES device including converters is assumed to be 100%, which is not realistic. 5. SFC data of MAK engine running @ 10% load is not available, which is assumed to be the same as Sulzer engine.

Specific fuel consumption 380 sfc1 vs. load1 360 sfc NO.1-6 DE (MAK 8M552) 340 sfc2 vs. load2 sfc NO.7-8 (Sulzer 8ZAL40S) and 9-12 DE (Sulzer 6ZAL40S) 320

300

280

260 SFC[g/kWh] 240

220

200

180 0 10 20 30 40 50 60 70 80 90 100 110 Load[% of MCR] Figure 4-13 Specific fuel consumption curves

As shown in Figure 4-7 in 4.2.1 engine load in existing power system and hybrid power system is calculated already from recorded load consumption data. In this way, fuel saving potentials of hybrid system by replacing one DG in each engine room with EES device in existing onboard power system can be estimated and results are outlined in

41 | Page

Table 4-10. It can be seen in the table that around 13% fuel savings can be obtained in DP and heavy lifting operation.

Table 4-10 Fuel consumption and saving potentials obtained from the Excel Model Engine room FC (kg) FC-hybrid (kg) Fuel savings (kg) Fuel savings (%) AFT (day1) 7553 5629 1924 25.47 PS (day1) 12865 11375 1490 11.59 SB (day1) 12880 11478 1402 10.89 Overall (day1) 33298 28482 4816 14.46 AFT (day2) 9382 7255 2127 22.67 PS (day2) 15857 14311 1546 9.74 SB (day2) 15341 13737 1604 10.45 Overall (day2) 40579 35304 5275 13

It should be noted that:

1. Fuel consumption calculated in the Excel model is still lower than “measurements/baseline” onboard with deviation of 12.69% and 9.86% for day1 and day2 respectively (not shown in Table 4-10), which can be induced by several reasons (corresponding to assumptions above) and will be tackled in detail in 6.3:

a. Fuel consumption measurements for the whole day (24 hours) and whole vessel (DE(s) and auxiliary systems) are obtained from readings of fuel tank. And fuel consumptions of DE(s) for the recorded periods are obtained by scaling the readings. Thus, “measurements/baseline” itself, is a roughly estimated value, which is assumed to be good enough to use in the concept design phase in this work.

b. Measurements of SFC data are based on factory trials decades ago, it is not clear that whether SFC curves in Figure 4-13 is good enough to predict the current DE fuel consumption characteristics. What’s more, test data in low load area (<25%) for MAK engine is not available.

c. Excel model on fuel consumption is an algebraic fit through measurements during DE trials, which is good enough for steady state and quasi-static operation of DE. However, it fails to account for impact of dynamic transient operation on fuel consumption, which might be the reason of deviations as well.

2. Results in Table 4-10 is based on replacing one DG in each engine room. If two DGs can be replaced by EES system in PS and SB engine room, more fuel savings can be achieved. Definitely, replacing two DGs in PS and SB engine room thereby requires a larger EES energy capacity which can be estimated in the same way as in Chapter 4. But this is not the original intention of this project thereby is out of the scope of this thesis. Therefore, details of replacing two DGs

42 | Page

will not be further explored in the thesis.

4.4 Synthesis

In this chapter, load consumption characteristics during DP mode is studied, which illustrates the opportunity for EES installed onboard to improve the efficiency of power system. What’s more, important technical parameters of EES device was assessed: energy capacity, (dis)charge power and full cycles (cumulative energy throughput). It should be noted the energy capacity of EES is estimated according to the idea of how “small” the capacity could be instead of how “big” the system should be. “Small” means a comprehensive offer which is a solution with enough capacity that aren’t over- dimensioned (“big”) to realize its defined functions. What’s more, both typical and top configurations of EES are derived from typical operational profile and assumed extreme scenarios that can be encountered. At last, efficiency improvement of power system through hybridization is verified by estimating fuel (MDO) saving potentials in Excel model. Deviations of fuel consumption figure and weakness of Excel model are also summarized and analyzed. 5. Electric energy storage (EES)

In fact, as summarized in Chapter 4.5, the initial estimation on EES configuration respectively for 3 designed functions is finished. However, it is based on the assumption of an ideal EES device. Therefore, this chapter provides a foundation for further discussion on which EES technologies, technically, might be most promising alternatives for Thialf hybrid system application tackling mainly three research sub-questions:

• What is the (approximated) EES configuration for Thialf considering EES energy efficiency and SOC regardless of different technologies? • Which current technologies or combination of technologies can be the alternatives for Thialf using Ragone plot and considering technical characteristics of different technologies? • What is the principle and state of development in terms of selected alternatives?

5.1 Configuration

5.1.1 Energy efficiency Firstly, it should be noted that although the kWh capacity is the measure of the energy content of the EES device. However, in battery industry, the “energy content” of a battery

43 | Page is usually expressed indirectly and “confusingly” by its electric charge capacity in A·h (Ampere-hour), without taking account of the battery voltage. The A·h (Ampere-hour) capacity denotes the total charge that can be discharged from a battery under specified conditions which is predefined by manufacture like in 20 °C, at a constant rate over a specified length of time. To convert kWh to A·h the kWh value must be divided by the voltage of the power source as shown by Equation 5-1. This value is approximate since the voltage is not constant during discharge of a battery. Therefore, the energy content of a battery depends on the battery voltage14 and this is different for different cell chemistries. ( ) = ( ) × ( ) Equation 5-1

Secondly,𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊ℎ𝑜𝑜𝑜𝑜𝑜𝑜 there𝑊𝑊ℎ are𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 several ways𝑉𝑉 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎to express− ℎ the𝑜𝑜𝑜𝑜𝑜𝑜 efficiency𝐴𝐴ℎ in battery industry, of which coulomb efficiency and energy efficiency is the most usually used for evaluating battery performance. The coulomb efficiency is defined in USABC (1996) as the ratio of the discharged capacity to the capacity needed to be charged to the initial state before discharge. The coulomb efficiency is shown in reference as [30]:

= 𝑡𝑡𝑑𝑑 Equation 5-2 ∫0 𝐼𝐼𝑑𝑑𝑑𝑑𝑑𝑑 𝐶𝐶 𝑡𝑡𝑐𝑐 𝜂𝜂Where ∫0 𝐼𝐼𝑐𝑐is𝑑𝑑 𝑑𝑑the discharge current, is the discharge time, is the charge current and is the charge time. But only the average coulomb efficiency in the whole charge 𝐼𝐼𝑑𝑑 𝑡𝑡𝑑𝑑 𝐼𝐼𝑐𝑐 and discharge processed are obtained from Equation 5-2 because experiment indicates 𝑡𝑡𝑐𝑐 that coulomb efficiency is different with respect to different current rates, SOC and temperature [31]. Definition of energy efficiency is similar with the coulomb efficiency. It defined as the ration of the discharge energy to the energy needed to be charged to the initial state before discharge. The energy efficiency is shown as:

𝑡𝑡𝑑𝑑 ∫0 𝑈𝑈𝑑𝑑𝑑𝑑𝑑𝑑 = 𝑡𝑡𝑑𝑑 = = = Equation 5-3 ∫0 𝑈𝑈𝑑𝑑𝐼𝐼𝑑𝑑𝑑𝑑𝑑𝑑 𝑡𝑡𝑑𝑑 �𝑈𝑈��𝑑𝑑� 𝑡𝑡 𝑡𝑡 𝑊𝑊 𝑐𝑐 𝐶𝐶 ∫ 𝑐𝑐 𝑈𝑈 𝑑𝑑𝑑𝑑 𝐶𝐶 𝐶𝐶 𝑉𝑉 𝜂𝜂 ∫0 𝑈𝑈𝑐𝑐𝐼𝐼𝑐𝑐𝑑𝑑𝑑𝑑 𝜂𝜂 0 𝑐𝑐 𝜂𝜂 𝑈𝑈��𝑐𝑐� 𝜂𝜂 𝜂𝜂 Where is the discharge𝑡𝑡 current,𝑐𝑐 is the discharge time, is the discharge voltage, is the charge current, is the charge time and is the charge voltage. 𝐼𝐼𝑑𝑑 𝑡𝑡𝑑𝑑 𝑈𝑈𝑑𝑑 and are functions of time. Equation 5-3 expresses the average energy efficiency in 𝐼𝐼𝑐𝑐 𝑡𝑡𝑐𝑐 𝑈𝑈𝑐𝑐 𝑈𝑈𝑑𝑑 the whole charge and discharge processes. Compared with coulomb efficiency the 𝑈𝑈𝑐𝑐 energy efficiency is more comprehensive since the voltage efficiency15 is also introduced in the definition, which is influenced by current rate, SOC, temperature and 𝜂𝜂𝑉𝑉 general conditions. Thirdly, it is known that key factors influence the energy efficiency is

14 The higher voltage battery, the charge is stored at a higher potential. For example, two batteries have the same A·h capacity means they contains the same charge in coulombs, but they may store different amounts of energy because of different battery voltage. It is like two identical containers both containing the same quantity of water, but one contains water at atmospheric pressure and the other contains water under high pressure. 15 Even a rechargeable battery with 100% coulomb efficiency requires charging at a higher voltage than it produces during discharge, so voltage efficiency below 100% reflect the thermodynamic irreversibility of every real-world chemical reaction.

44 | Page

(dis)charge rate, SOC, temperature and internal resistance.

Therefore, in terms of initial calculation of EES configuration like energy capacity, energy efficiency of different EES devices should be considered respectively and it is shown in Appendix C. But in order to simplify the initial calculation process at this phase, an initial proposed topology is used here (illustrated in Figure 5-2) to present the electric energy transformation between energy storage device and the high voltage ac grid as presented in Figure 3-7. It should be mentioned that the DC/DC converter is shown by a dashed line box because it is not essential for all of the EES technologies. For example, there is no need to install the dc/dc converter for a battery technology with a flat discharge curve, because it can deliver 90 to 95 percent of its energy reserve before reaching the voltage threshold. However, for example, taking a 6V power source that is allowed to discharge to 4.5V before the equipment cuts off. For supercapacitors with the linear (dis)charge, the supercapacitor reaches the voltage threshold with in the first quarter of the cycle and the remaining three-quarters of the energy reserve become unusable. Therefore, the dc/dc converter can be used to maintain the power level by drawing higher current with dropping voltage, but this adds costs and introduces loss [32].

Losses of converters are due to heat development in the material; copper losses (internal resistance) and voltage drop at the diodes and thyristors. These losses are load dependent, next to this there is usually a no-load loss in the form of a running cooling fan. [33] Mentions the distribution between these three to be 45% conduction (IR), 45% stitching losses (I2R) and 10% steady losses:

, = (1 ) (0.1 + 0.45I + 0.45( ) ) ∗ ∗ 2 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑛𝑛𝑛𝑛𝑛𝑛 𝑛𝑛𝑛𝑛𝑛𝑛 Converter manufactures𝑃𝑃 all specify𝑃𝑃 a∙ nominal− 𝜂𝜂 efficiency∙ between 0.97 𝐼𝐼and 0.98. For a constant voltage the normalized current I is proportional to normalized power P , so according to it can be stated that the efficiency∗ will drop for low loads. However, the∗ range between 10% and 100% load is within 1% of nominal efficiency for a nominal efficiency of 98%, see Figure 5-1 [34].

Efficiency FC Eta FC 0,99 0,98 0,97 0,96 0,95 0,94 0,93 0,92 0,91 0,90 0,0 0,2 0,4 0,6 0,8 1,0 1,2 ∗ 푷

Figure 5-1 Efficiency of frequency converter (Source:[34])

45 | Page

In order to simplify the general calculations in this phase, a fixed efficiency for converters is assumed. And also for the EES device and transformer. It is assumed that (round-trip) energy efficiency at nominal point for EES device is 95%, meanwhile efficiency of DC- AC/AC-DC inverters is 97% and efficiency of transformer is 98%. Thereby, the round trip energy efficiency of EES system connected to onboard AC grid (AC-AC efficiency) is 80.76% (98%*97%*97%*95%*97%*97%*98%) as illustrated in Figure 5-2.

92.2*0.95 87.59*0.97 84.96*0.97 82.41*0.98 =87.59 kWh =84.96 kWh =82.41 kWh =80.76 kWh

HV EES DC/DC DC/AC Transformer AC Grid

92.2 kWh 95.06 kWh 98 kWh 100 kWh = 95.06*0.97 = 98*0.97 =100*0.98

Figure 5-2 Example of AC-AC efficiency of EES connected to HV AC grid

From Table 4-5 it can be concluded that 78 kWh energy is required for EES device in order to realize enhanced dynamic performance function in daily typical DP scenario. According to Table 4-3, it can be found that 230 kWh energy should be contained in EES device to ensure enough capacity is always left for spinning reserve function in case of one DG fails at any time. Thereby energy capacity of EES device can be calculated as shown in Equation 5-4 in order to realize both of two functions mentioned above, although this strategy is conservative.

= = (78 + 230)/(0.98 0.97 0.97 0.95) = 344

𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐸𝐸𝐸𝐸Equation𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑐𝑐5𝑐𝑐𝑐𝑐𝑐𝑐-4 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 ∗ ∗ ∗ √ 𝑘𝑘𝑘𝑘ℎ It should be noted that 344 kWh energy capacity is also enough for EES to fulfill enhanced dynamic performance function in most extreme operation mentioned in sub- chapter 4.2.5. Thereby conservative energy capacity16 for EES can be defined as EES can afford one DG fails for 5 minutes immediately after enhanced dynamic performance function has been performed during full load crane operation, which is 486 kWh. For practical reasons, the possibility of this situation is quite low and can be handled by other methods. Therefore, 344 kWh is the value of EES energy capacity given the assumption of AC-AC energy efficiency is 80.76%.

5.1.2 SOC and DOD In different papers [35-39] different definitions for SOC were used and methods for SOC estimation were also described. In order to understand what the term “SOC” really represents, a clear definition in the simplest way is used in [39]: the ratio between the current (A·h) capacity ( ) and the nominal (A·h) capacity ( ) which is given by manufactures and represents the maximum amount of charge presented in the fully 𝑄𝑄𝑐𝑐 𝑄𝑄𝑛𝑛

16 16 = (206 + 230)/(0.98 0.97 0.97 0.95) = 486

𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑒𝑒𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 ∗ ∗ ∗ √ 46𝑘𝑘 |𝑘𝑘 Pageℎ charged new (not the current) battery (Equation 5-5).

SOC = Equation 5-5 𝑄𝑄𝑐𝑐 Depth of𝑄𝑄 discharge𝑛𝑛 (DOD) is an alternate complement method to indicate SOC: the ratio between discharged capacity and nominal capacity. Usually a discharge to at least 80% DOD is referred to as a deep discharge.

DOD = = 1 SOC Equation 5-6 𝑄𝑄𝑑𝑑 𝑄𝑄𝑛𝑛 − In order to simplify calculation in this chapter it is assumed the battery voltage is fixed and the charge stored is directly proportional to energy stored so that kWh capacity can be used to calculate SOC and DOD instead of A·h capacity. In terms of 344 kWh EES device mentioned in 5.1.1, a large part of energy (257 = 230/(0.98 0.97 0.97 0.95) is exclusively stored for spinning reserve function which will be rarely used in case 𝑘𝑘𝑘𝑘ℎ ∗ ∗ ∗ of one DG fails. And a small percentage (87 = 78/(0.98 0.97 0.97 0.95) is √ almost continuously used for enhanced dynamics performance function as shown in 𝑘𝑘𝑘𝑘ℎ ∗ ∗ ∗ √ pervious chapter and also Figure 0-4 in Appendix A. Thereby SOC and DOD can be calculated:

= 𝑡𝑡𝑑𝑑 100%, 100% = [74.7%, 100%] 𝑄𝑄𝑛𝑛𝑈𝑈𝑛𝑛−∫0 𝐼𝐼𝑑𝑑𝑈𝑈𝑑𝑑𝑑𝑑𝑑𝑑 344−87 344 Equation𝑆𝑆𝑆𝑆𝑆𝑆 5-7 𝑄𝑄𝑛𝑛𝑈𝑈𝑛𝑛 ∈ � 344 ∗ 344 ∗ � ( ) . . . . = 78−50 100% = 90.9% Equation 5-8 344−0 98∗0 97∗0 97∗√0 95 = 100% 74.7% = 25.3% Equation 5-9 𝑆𝑆𝑆𝑆𝑆𝑆𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 344 ∗ Where is the nominal capacity, is the nominal voltage, is the discharge 𝑚𝑚𝑚𝑚𝑚𝑚 𝐷𝐷voltage,𝐷𝐷𝐷𝐷 is the charge− current, is the discharge time. Equation 5-7 and Equation 𝑄𝑄𝑛𝑛 𝑈𝑈𝑛𝑛 𝑈𝑈𝑑𝑑 5-9 shows that upper and lower boundary of SOC is respectively 100% and 74.7%, in the 𝐼𝐼𝑑𝑑 𝑡𝑡𝑑𝑑 meantime, SOC is around 90.9% most of times. What’s more, the maximum DOD of 25.3% actually is rarely been achieved and most of times EES is exposed to very shallow cycles as shown in Appendix A Figure 0-4 which is beneficial for cycle life as well. According to table in Appendix B, maximum 25.3% DOD can be achieved in terms of most EES technologies without degraded performance. Thereby, it can be concluded that there is no need to increase EES energy capacity (344 kWh) after taking SOC and DOD into account.

5.1.3 C-Rate The function of EES device is realized by receiving and releasing current (energy) at the desired time and in a controlled manner. Today, battery industry uses C-Rate to express the current handling capability of a battery. C-Rate scales the current at which a battery is discharged or charged in order to normalize against battery A·h capacity. The (dis)charge rate of a battery is expressed as C/R, where R is the number of hours required to completely discharge its nominal capacity [28]. For example, if a battery has a nominal capacity of 60 Ah, discharge at the rate of C/20 would fully discharge it in 20h thereby the

47 | Page current is 3 A. And if the discharge rate is C/5 the discharge current is 12 A17 [28]. The C- Rate is introduced here in this work because for Thialf’s application it is essentially important to use the smallest energy EES capable of supplying the required power.

This definition can be approached from two sides: One, it can be defined as the current. Two, it can also defined as the time it takes for a battery to discharge. But the constant power discharge mode is becoming more popular for battery-powered applications and it is also the case for Thialf EES application18. A method, analogous to the C rate, can be used to express the discharge or charge rate in terms of power: E-Rate [27]. For example, the power level at the 0.5E rate for a battery rated at 12Wh is 6W and it takes the battery 2 hours to completely discharge 12Wh energy. Therefore, in the same way, this definition can be approached from two sides as well: One, it can be defined as the power. Two, it can also defined as the time it takes for a battery to discharge, which is the same as the C-Rate. Basically, according to the definition we can find that for a very flat EES discharge curve the result from C-Rate and E-Rate is almost consistent with each other because the voltage is fairly constant during most duration of discharge like Li-ion battery. But for a more sloping discharge profile the C-Rates value is higher than the E- Rate value because of the decreasing voltage during discharge.

In order to simplify the calculation here no distinction of the two definitions is made mostly in this thesis. And the term “C-Rates” is always used in the thesis (not the E- Rates) from here.

Since energy capacity of 344 kWh is obtained after previous analysis and EES (dis)charge power can be obtained from Table 4-6, Table 4-3 and Table 4-8. Therefore, “R” can be derived from energy-to-power ratio, which means C-Rate is obtained. Statistical data of C-Rate is shown in Table 5-1 and Table 5-2. It can be seen in Table 5-1 that in terms of typical peak shaving function, the average (dis)charge C-Rate of EES is quite low at around 0.3C but the maximum C-Rate can achieves considerably high for example at 9.1C discharge for one second. Different EES technologies afford C-Rates which varies considerably, some provide continuous low C-Rate for a long time while some are used to offer far more for a shorter duration. Thereby, C-Rate expected from EES device installed on Thialf in terms of different designed functions can be used as an important specification for choosing the right EES technology.

17 It is to be noted that the capacity of a battery generally decreases with increasing discharge current. Thus the battery rated at 5 Ah at the C/5 rate (or 1 A) will operate for 5 h when discharged at 1 A. If the battery is discharged at a lower rate, for example the C /10 rate (or 0.5 A), it will run for more than 10 h and deliver more than 5 Ah of capacity. Conversely, when discharged at its C rate (or 5 A), the battery will run for less than 1 h and deliver less than 5 Ah of capacity. 18 The current increases during the discharge as the battery voltage decreases, thus discharging the battery at constant power level. (More discharge mode is available in 27. Linden, D., Linden's Handbook of Batteries. 4th ed, ed. T. B.REDDY. 2010.)

48 | Page

Table 5-1 C-Rate required from EES for typical peak shaving function EES discharge Counts Counts Counts Counts Counts Counts C-Rate PS day1 SB day1 PS day2 SB day2 AFT day1 AFT day2 < 1C 19510 16251 19298 17077 21543 26438 1C-2C 822 747 941 891 668 622 2C-3C 159 228 290 260 275 350 3C-4C 81 47 123 58 131 217 4C-5C 19 12 80 29 12 142 5C-6C 4 8 76 19 10 45 6C-7C 2 2 32 0 3 6 7C-8C 0 0 7 0 5 0 8C-9C 0 0 6 0 0 0 9C-10C 0 0 1 0 0 0 Avg. C-Rate 0.19C 0.22C 0.30C 0.26C 0.19C 0.23C Max. C-Rate 6.8C 6.8C 9.1C 5.9C 7.5C 6.8C EES charge Counts Counts Counts Counts Counts Counts C-rate PS day1 SB day1 PS day2 SB day2 AFT day1 AFT day2 < 1C 21041 26669 27616 33640 25423 29542 1C-2C 768 748 777 851 825 1201 2C-3C 172 81 154 55 89 214 3C-4C 29 19 110 0 0 78 4C-5C 9 16 162 0 0 0 5C-6C 18 16 55 0 0 0 6C-7C 17 20 27 0 0 0 7C-8C 0 0 10 0 0 0 Avg. C-Rate 0.18C 0.14C 0.22C 0.14C 0.17C 0.21C Max. C-Rate 6.7C 6.6C 7.5C 3.0C 2.6C 3.7C

Table 5-2 C-Rate required from EES for spinning reserve function and demanding peak shaving function

C-Rate Spinning reserve function Demanding peak shaving function Avg. C-Rate 9C discharge, 5C charge 9C discharge and charge Max. C-Rate 12C discharge, 6C charge 17C discharge and charge

However, according to Table 5-2, in case of DG’s failure, spinning reserve function is activated, which means EES is designed to discharge at 9C and charge at 5C continuously and endure discharge pulse as high as 12C. What’s more, demanding peak shaving function requires a much more challenging 9C average and 17C maximum (dis)charge in most extreme operations.

But it also should be noted that both spinning reserve function and peak shaving function

49 | Page in such a harsh instance would be useful just in extremely rare circumstances. That is to say, in terms of (dis)charge rate Table 5-2 sets the technical ceiling for EES and Table 5-1 sets the typical operational profile for EES as shown in Figure 5-3.

Table 5-2 (Spinning reserve function)

Table 5-2 (Demanding peak shaving function) Table 5-1

Figure 5-3 Border stands for EES function

What’s more, charge rate shown in Table 5-2 actually is not an essential design criteria because on one hand the charge rate can be controlled by the PMS easily (Figure 4-6 and Figure 4-12) to endure the EES life cycles, on the other hand the 9C continuous fast charge rate capability close the door for most battery technologies, otherwise, it makes a battery system over dimensioned with a very large capacity which can be seldom used. But a fast charge capability of EES can be a strength because it helps to ensure smooth operation of DE19. Thus, a design criteria for Thialf in terms of C-Rates can be proposed and listed in Table 5-3. The nominal value is picked from the maximum average value in Table 5-1. The max continuous discharge is picked from Table 5-2. And the max continuous charge is picked form Table 5-2 of the spinning reserve function, although this is a very conservative strategy because as mentioned above this value can be reduced. The max value discharge is picked from Table 5-2 enhanced dynamics performance function. And the max charge value is picked from Table 5-1 the max value of max charge C-Rate.

Table 5-3 C-Rate design criteria for Thialf C-Rate discharge charge Nominal 0.3C 0.22C Max continuous 9C 5C20 Max 17C 7.5C

19 In case the fast charging is problematic for some types of batteries, resistors could also be installed for burning the energy, which is used presently for the regenerated energy of cranes, rather than using the energy for charging the batteries. 20 Continuous 5C charge capability is not a necessary condition. But EES with such a capability will be favored.

50 | Page

5.2 Ragone plot

Before comprehensive analysis and comparison of different EES technologies, an initial and primary comparison of them can be made by using Ragone plot which is given in Figure 5-5. But before introducing Ragone plot, EES classification is introduced briefly.

5.2.1 EES classification A widely-used approach for classifying EES systems is the determination according to the form of energy used [40]. Some of the key technologies, not all of which are at the stage of commercial application are shown in Figure 5-4.

What’s more, technical data for the different technologies plotted in Figure 5-5 is according to Table 0-1 in Appendix F that provides the matrices to clearly show the positions of different EES performance and characteristics, which is based on both academic research reports and commercially available experience and is taken from comprehensive literature review during the project, sources and webpages like [24, 27, 40-45].

Figure 5-4 Classification of electrical energy storage technologies according to energy form (Source:[40])

Related technologies and abbreviations are listed below:

• EES store electric energy as mechanical energy like PHS: pumped hydro storage, CAES: compressed air energy storage, FES: flywheel energy storage;

• EES store electric energy as chemical energy like LA: lead acid battery, NaS: sodium Sulphur battery, NaNiCl: sodium nickel chloride battery, NiCd: nickel cadmium battery, NiMH: nickel metal hydride battery, Li-ion: lithium-ion batteries, VRFB: vanadium redox flow battery, HFB: hybrid flow battery;

• EES store electric energy directly like SC: supercapacitor, SMEC: superconductiong magnetic energy storage.

51 | Page

5.2.2 Ragone plot The Ragone plot originated in a historically significant paper by David V. Ragone at the May 1968 meeting of the Society of Automotive Engineers at Detoit, MI, where the first examples were shown to compare performance of batteries and thus named after him. Today Ragone plot is used for performance comparison of various energy-storing devices. As can be seen in Figure 5-5, the values of energy density (Wh/l) are plotted versus power density (W/l) and presented in a log–log space, which allows comparing performance of very different devices like extremely low and high power density. The (dis)charge time of an application is of the order of the energy-to-power ratio of the EES device and in the log–log plane of Figure 5-5, the time corresponds to straight lines with unit slope. In addition, Figure 5-5 is presented on a basis which allows a general initial comparison of different technologies and thus the plot are focus on the specific regions where Ragone curves are located rather than the specific curves21 [46, 47].

Ragone plot 10 3 Li-ion NaS NaNiCl NiMH 2 LA 10 HFB

NiCd VRFB 1 10h FES 10 CAES

10 0 1h 6.6m 3.5m Energy Density[Wh/l] 36s 3.6s 1s

PHS -1 10 -1 1 2 4 5 10 10 0 10 10 10 3 10 10 Power Density[W/l]

Figure 5-5 Ragone plot

In Figure 5-5, for a given amount of energy, the higher the power and energy densities are, the smaller the volume of the required energy storage system will be. Therefore, highly compact technologies suitable for volume-limited applications can be found at the top right corner and the large volume consuming storage systems are located at the

21 So called Ragone curves are almost the same with Ragone plots but usually have a hooked shape which corresponding to a fall-off of energy density as power density (i.e. power drain) is increased in discharge of the battery instead of a rectangular shape in Figure 5-5. The general origin of it was recognized qualitatively in terms of increasing cell polarization (hence diminution of cell voltage which determines energy density) as higher power demands are made on a battery or capacitor at higher rates of discharge. Thus, Ragone curves not only provides the limit in the available energy and power of an EES device but also gives the optimal region of working, which is given by the part of the curve where both energy and power are high.

52 | Page bottom left corner. It can be easily seen that PHS and CAES have the lowest power densities and thus are mainly used in stationary EES system which require large reservoirs for grid scale and long duration applications from one hour to more than ten hours.

Most secondary batteries like LA, NaS, NaNiCl, NiCd and NiMH have relatively moderate energy densities and power densities, which offers a discharge time from minutes to hours. Therefore, they are used to serve different EES application domains. Li-ion have the best technical characteristics because of both high energy density and high power density, which offers reasonable explanation to the current broad range of development and applications especially in portable devices and transportation applications. Whereas, VRFB and HFB commonly have lower energy densities than those of conventional batteries. With other technical characteristics they are extremely suited to use in large fixed power applications for long periods from hours to principally more than 10 hours.

FES and SC have a relative higher power density, meanwhile, fast response time and long cycle life (Appendix B). Thus, they perform well for power quality applications of electric power (current) delivery for a short time from seconds to minutes.

In terms of discharge time and C-Rates, it can be seen in the plot that several straight lines are plotted with designations of discharge time. For example the 6.6 minute and 3.5 minute straight line in fact respectively stand for the 9C and 17C discharge line. EES technologies lies on the right side of 9C discharge line means they can deliver continuous 9C or higher C-Rate discharge. Thereby LA, NiCd, NiMH, Li-ion, FES and SC are on the list of alternative technologies after primary selection. It should be noted that SMES is excluded due to practical reasons like commercially availability and high operational cost.

5.3 Alternatives

Since there are already six competitors on the alternatives list: LA, NiCd, NiMH, Li-ion, FES and SC, thereby the next step is to pick out the most promising one or combination for Thialf through comparison. The selection of process relies on the assessment in the characteristics of different EES options against the performance requirements of application.

It should be noted that, as discussed in [27], these types of data and comparisons in this section (as well as performance characteristics in Appendix C) are necessarily approximations, with each system being presented under favorable discharge conditions. The specific performance of the EES device is very dependent on the cell design and all the detailed and specific conditions of the use and discharge-charge of the EES device.

5.3.1 Thialf application The prime requirement for EES for Thialf’s application is good enough current handling capability (C-rates), required life cycles and safety (see 9.1), if possible high energy

53 | Page density (which means less dimension, see 5.3.3), high efficiency, no maintenance requirements, and low cost (will be discussed in Chapter7):

• Current handling capability—as discussed in 5.1.3 both high discharge and charge rates from time to time are expected from EES technologies. According to 5.2.2, several technologies fulfilling discharge requirements (9C continuous) are selected as alternatives. However, charge (5C continuous) is more abusive for some technologies like batteries than discharge. Thereby, EES technologies can afford a maximum 5C continuous charge without degradation and safety problems will be favored.

• Cycle life—estimating EES life cycles accurately considering comprehensive factors which strongly influences the cycle life like temperature, DOD, (dis)charge rate and mechanical failure is extremely complex. Thereby in a very simplified way, the full equivalent cycles method is used here in the thesis. According to Equation 4-2 and Equation 5-9, the daily equivalent full (dis)charge cycles of EES installed on Thialf is around 3.9 per/day (1200 kWh/day/308 kWh). Therefore, if 10 years designed calendar life22 of EES is set for Thialf, then it requires full equivalent cycles of at least around 7600 (3.9*365*10*0.53). It should be noted that this target (7600 full equivalent cycles for 10 years) has been set under the typical (dis)charge conditions with SOC lies from 74.7% to 90.9% (summarized in 5.1.2) and average (dis)charge rate is around 0.3C (summarized in 5.1.3).

• Efficiency—since the main target of hybridization is fuel saving thereby a high efficiency EES device is preferred, which requires both high energy efficiency and low self-discharge23. Low energy efficiency of the EES system not only means high losses but also high costs for the compensation of these losses like extra cooling systems for dissipation of the generated heat. And self-discharge should be taken into consideration for spinning reserve function because EES acts like a UPS and seldom been used.

• Operation (maintenance) requirements—although extra EES device can reduce the maintenance workload of DGs but it will introduce extra maintenance work according to different EES systems. Thus EES devices with less maintenance requirements or free of maintenance are preferred.

5.3.2 EES technology comparison Since design requirements of EES system are defined in previous sections and summarized in 5.3.1 thereby a general comparison of alternative EES technologies can be made and shown in Table 5-4. As can be seen in the table:

• Charge rate—among battery technologies, only (some types of) Li-ion batteries

22 Calendar life is a separate issue from cycle life. Some failure modes are independent of charge-discharge cycling and are driven by different time-dependent chemical processes such as corrosion. 23 Loss of capacity (charge) when no external load is applied.

54 | Page

can afford such a rapid charge capability as high as continuous 5C. Figure 5-6 shows capacity retention versus cycle number for the AHR18700M1Ultra graphite/LiFePO cell from A123 Systems when the cell is charged and discharged at a 10C rate at room temperature. More than 93% capacity remains after 1400 cycles [27]. The common feature of FES and SC is the possibility of extremely high rate of discharge and charge, which means both of them are capable of handling 5C continuous charge. Therefore, in terms of charge rate, Li- ion, FES and SC are favored in terms of Thialf application.

+10C/–10C cycling at room temperature 100% +10C/–10C 90% 80% V , y

t 70% i c 60% ap a c 50% al i i t

i n 40% t

e n 30% c e r

P 20% 10% 0% 0 250 500 750 1000 1250 1500 1750 Cycles Figure 5-6 capacity retention versus cycle number for the AHR18700M1Ultra graphite/LiFePO cell from A123 Systems when the cell is charged and discharged at a 10C rate at room temperature. (Source: [27])

• Cycle life—a simple comparison on rating of cycle life (source: [27]) is given in Table 5-4 since specific performance is dependent on the particular design and conditions under which the EES is used. Appendix C also gives empirical data of approximate typical cycle life of different technologies. It can be concluded that only Li-ion, FES and SC can fulfill the lifecycle baseline requirement for Thialf application. But it should be also noted that because of the many possible interactions between different factors which influence the performance of EES technologies, these effects can be presented only as generalizations and that the influence of each factor is usually greater under the more stringent operating conditions [24]. Thus, the combined effects of different factors for Thialf application on lifecycles will be dealt with in more details in 6.5.5. • Efficiency—apparently Li-ion, FES and SC is superior to other types of EES technologies in terms of energy efficiency. Considering the self-discharge effect into account as well, Li-ion battery is the best then. • Maintenance—it is just a kind of rough indication here that several types of EES in general don’t require additional maintenance. But it should also be noticed that other technologies also have the non-maintenance design according to different manufactures and designs.

55 | Page

Table 5-4 EES technologies comparison EES LA NiCd NiMH Li-ion FES SC

Rapid charge 1C 1C24 1C 10C N/A N/A Rate

Recommended 0.07C 0.2C 0.1C <0.8C N/A N/A charge rate Cycle life25 4 3 3 2 1 1

Energy 70%-75% 60%-70% 55%-65% 95% 80%-95% 90%-94% efficiency26 93%-95% [27] 20% first 24 Up to Up to 25% Self- 10% hours, 2%-%10 20%/hour 5%/month in the first discharge27 /month 10%/month /month (100% 48 hours thereafter daily)

Maintenance Yes Yes No No Yes No

Nominal cell 2.0 1.2 1.2 2.5-4.2 N/A 2.3-2.75 voltage (V)

To sum up, in terms of Thialf onboard application; Li-ion, FES and SC can be good alternatives mainly because of acceptable enough lifecycles and good current handling capability within the designed energy capacity (344kWh).

5.3.3 Dimension analysis EES design usually take two directions; it is about trade-offs: either system will have lots of power and be used as a peak shaving power source, or lots of energy and they will be used as the main energy source for backup power, but one can’t generally have both. There is a huge difference in dimensioning the EES as a power source versus an energy source. However, for a total operation, like Thialf, there is a demand for both high power and high energy28 [8].

The specification of EES system which used for rough estimation of system dimension can be concluded from the previous chapters and outlined in Table 5-5: Table 5-5 Specification of EES Specifications Value Energy capacity ≈ 344 kWh Power capacity (Continuous discharge) ≈ 3096 kW

24 Fast charge designed NiCd batteries. 25 Dependent on DOD. Rating: 1 to 5, best to poorest. 26 All date are related to normal rates of charge and discharge and room-temperature operation. 27 Loss of capacity [charge] when no external load is applied. 28 For demanding enhanced dynamic performance function and spinning reserve function.

56 | Page

Dimension analysis is designed according to energy and power requirements in Table 5-5 and based on energy and power density in Table 0-1 in Appendix C.

But it should be noted that values of energy density (Wh/l), specific energy (Wh/kg), power density (W/l) and specific power (W/kg) which used for dimensional analysis outlined in Table 0-1 from different literatures are not always align with each other. Actually sometimes, the deviation between different sources is even as large as one order of magnitude.

On one hand, this is due to values in the table are picked out from different sources and they are sometimes not indicated as theoretical or practical values in the original reference. For example, in terms of batteries, the theoretical capacity in ampere- hours/gram (Ah/g) is based on the equivalent weight of the active materials participating in the electrochemical reaction. Multiplying the theoretical capacity and voltage gives the theoretical specific energy in Wh/kg [48]. However, practical systems in real life do not achieve these maximum theoretical values. One obvious reason is that a practical in practice battery has a number of passive components that are not involved in the basic chemical reaction that acts as the energy storage mechanism. They add to weight and volume, but do not contribute to the transduction between electrical and chemical energy [28]. A rule of thumb that was used for a number of conventional aqueous electrolyte battery systems in the past was that a practical cell could only produce about 1/5– 1/4 of the maximum theoretical specific energy. But optimization of a number of factors has made it possible now to exceed such values in a number of cases [28]. For FES system some literatures give the energy density or specific energy based on the volume or mass of only the rotor. On the other hand, commercially available values from the website of manufactures can be different from the values which has only been achieved academically or in the laboratory. The last but not the lease, EES technologies have been developing rapidly which leads to deviation of different publications because the maximum theoretical values or practical values of some of the newer electrochemical systems are considerably higher than which can be available earlier.

Considering the objective of this project is to looking for opportunities among mature developed commercially available EES technologies. Thus, in order to determine the practical-oriented values for determining alternatives for Thialf and also for calculating dimension of EES device more accurately, a survey of prominent manufactures of different technologies has done to verify/modify different values from different sources which please refer to Appendix D and E. And finally, practical values instead of theoretical values can be determined.

Li-ion battery dimension Table 5-10 outlines the general performance characteristics of Li-ion batteries. Cells having a specific energy up to 240 Wh/kg and energy density up to 640 Wh/l are commercially available[27]. But due to high power requirement of the installation onboard, specifications of “power cells” (see Table 5-6) are used here to estimate dimension of cells, which is fairly realistic.

57 | Page

Table 5-6 Li-ion power cell general specification Li-ion Power cells specifications Value Energy density (Wh/l) 120-350 Specific energy (Wh/kg) 60-150 Power density (W/l) 2000-10000 Specific power (W/kg) 1100-4000

Dimension result is outlined in Table 5-7 below. But it should be noticed that the volume and weight estimations in Table 5-7 is only for Li-ion battery cells. And even if the lowest value of energy density and specific energy is selected, the weight and volume of the Li- ion battery cells is 5.8 ton and 2.9 m3.

Table 5-7 Dimension of Li-ion battery Dimension Value Volume (energy requirement) 983-2867 L Weight (energy requirement) 2294-5734 kg Volume (power requirement) 310-1548 L Weight (power requirement) 774-2815 kg FES dimension The method use to derive practical specifications for FES device is shown in Appendix E. Thus in the same method, dimension results for FES is shown Table 5-8. And if a typical value of 5Wh/kg and 10 Wh/l is used, thereby the weight and dimension of FES modules is 68.8 ton and 34.4 m3.

Table 5-8 Dimension of FES Dimension Value Volume (energy requirement) 17200-344000 L Weight (energy requirement) 34400-344000 kg Volume (power requirement) 620-5160 L Weight (power requirement) 2064-7740 kg SC dimension

The method use to derive practical specifications of SC device is shown in Appendix D. Also in the same method, dimension results for SC is shown in Table 5-9. And if a typical value of 4Wh/kg and 3 Wh/l is used, thereby the weight and dimension of SC modules is 86 ton and 115 m3.

Table 5-9 Dimension of SC Dimension Value Volume (energy requirement) 57334-344000 L Weight (energy requirement) 22934-344000 kg Volume (power requirement) 155-15480 L Weight (power requirement) 207-1548 kg

58 | Page

5.3.4 Li-ion The term “Lithium-ion” is used generically for many products, but the details of the architecture and chemistry imply very different performance characteristics for batteries since Lithium-ion batteries have a large number of electrolytes and combinations of electrodes materials [24, 49]. The general performance characteristics of Li-ion batteries are outlined in Table 5-10. As indicated in the table, single cells typically operate at a high voltage between 2.5 and 4.3 V, which is approximately three times that of NiCd or NiMH cells, and thus fewer cells are required for a battery pack of a given voltage.

Another important finding is power density and energy density is generally inversely proportional. Survey from DNV GL [49] represents a significant cross section of the present-day industry and determines that with specific energy greater than 150 Wh/kg, it is unlikely that the battery is capable of specific power greater than 2000 W/kg. However, for lower energy densities, some chemistries can offer up to 12000 W/kg. Therefore, it is common for manufactures to offer “energy cells,” where specific energy and energy density are maximized, and “power cells,” where specific power and power density are maximized while still retaining energy density greater than competitive technologies29.

29 Cell design and its chemistry which has an impact on its ability to deliver high power, which should be taken into account according to specific applications in different environmental conditions and engineering requirements. Thereby it should not be said that a Li-ion battery designed for portable electronics is necessarily scalable to large-format, more-extreme power systems like in marine sector.

59 | Page

Table 5-10 General performance characteristics of Li-ion cells (cylindrical, prismatic, and "polymer") using common cell chemistries (Source:[27]) LCO/graphite NMC/graphite NMC/graphite LFP/graphite Characteristic LMO/graphite LMO/LTO 30 NCA/graphite Power cells Power cells Energy cells

2.5-4.2 typ. 2.5- Voltage range (V) 2.5-4.2 2.5-3.6 2.8-1.5 4.35 for some cells

Avg. Voltage 3.7 3.7 3.3 2.3

Energy density (Wh/l) 400-640 cylinder 250-350 125-250 120 250-450 polymer

Specific energy (Wh/kg) 175-240 cylinder 100-150 60-110 70 130-200 polymer

Power density (W/l) 2000 10000 10000 2000 (pulse)

Specific power (W/kg) 1000 4000 4000 1100 (pulse)

Continuous C-Rate 2-3 Over 30 10-125 10

Pulse C-Rate 5 Over 100 Up to 250 20

Cycle life at 100% DOD (to 500+ 500+ (1000+) (4000+) 80% capacity)

Charge temperature range 0-45 0-45 0-45 -20-45 (°C)31

Discharge temperature -20-60 -30-60 -30-60 -30-60 range (°C)

For example, Tesla motors use Panasonic 18650 NCA “energy cells” on Model S because of high specific energy of 160 Wh/kg. One of the “power cell” examples is a specifically designed, extremely high-power, graphite/LFP cell (the AHR18700M1Ultra), which has been produced by A123 system and successfully used by Vodafone-McLaren- Mercedes Formula 1 team in Kinetic Energy Recovery System (KERS) throughout the 2009 racing season. This high-power cell is designed for use at operating temperatures up to 100°C, can deliver specific power of over 20 kW/kg, and provides specific energy of 60 Wh/kg. Under race conditions, the discharge C-Rate is about 250C, corresponding to about 80% of the stored energy being delivered in 6 to 8 seconds.

30 Cathode: LMO, anode: LTO. 31 Some cells have wider range.

60 | Page

From the table and other technical characteristics it can be also concluded that LTO battery is extremely suit for Thialf application, considering reasons below:

1. Best cycle life performance among high-power Li-ion families (more calculations and analysis, refer to 6.5.5),

2. Symmetric current handling capability for both discharging and charging process,

3. High levels of safety and reliability e.g., without prone to fire or explosion even from internal short circuit caused by external pressure.

5.3.5 FES The flywheel is an old means of storing energy and smoothing out power variations. The potter’s wheel and spinning wheel are examples of historical uses of flywheels. Energy can be transferred into the flywheel either electrically or through a mechanical connection. The current trend is towards electric machines, transferring power in a contact-less manner between rotor and stator. This introduces two important advantages: a long lifetime and low standby losses [50].

Electrical-transferred FES works by, during charging process, a motor is used for exerting a positive torque and accelerating the rotor (flywheel) to a very high speed and maintaining the energy in the system as rotational energy. During discharging the kinetic energy is extracted by a generator (applies a negative torque) driven by the inertia of the rotor (flywheel) resulting in a deceleration of the rotating mass. The usable kinetic energy stored in a flywheel is the sped interval over which it is allowed to operate:

1 E = = 2 2 2 2 2 𝜔𝜔𝑚𝑚𝑚𝑚𝑚𝑚 − 𝜔𝜔𝑚𝑚𝑚𝑚𝑚𝑚 2 𝐽𝐽∆𝜔𝜔 � 𝑟𝑟 𝑑𝑑𝑑𝑑 Where is the moment of inertia about the axis of rotation; is the rotational velocity and dm is a small mass element at a distance r from the axis of rotation[50]. It can be 𝐽𝐽 𝜔𝜔 concluded that more energy is attained in a linear fashion by adding extra weight and the kinetic energy grows quadratically with radius and rotational speed. Thus, advanced FES systems have rotors made of high strength carbon-fiber composites which can spin at speeds from 20,000 to over 50,000 rpm using magnetic bearings a vacuum enclosure. Figure 5-7 shows main components of a typical flywheel.

61 | Page

Figure 5-7 Main component of FES

The main features of flywheels are high charge and discharge rates for many cycles, little maintenance, wider operation temperature range and no DOD effects (the cycle life of flywheels is not directly related to DOD). The main weakness of FES is that ﬂywheel devices suffer from the idling losses during the time when the ﬂywheel is on standby. This can lead to relatively high self-discharge, up to 20% of stored capacity per hour [44, 51].

FES can be used in buses and cars for power boost and regenerative braking (usually through mechanical power transfer). FES is also becoming a mature technology for grid power applications by combining multiple units into a “flywheel farm”. A 5MWh (20 MW over 15 minutes for frequency regulation) flywheel energy storage plant in New York which comprises 200 high-speed high energy flywheels was put into service in 2011. In this market, flywheels performed between 3,000 and 5,000 full depth-of-discharge cycles a year.

5.3.6 SC SC are a high-capacity electrochemical capacitor with capacitance values up to 10,000 farads which fill the gap between electrolytic capacitors and batteries. SC are basically double-layered versions of normal capacitors but with considerably higher electrode surfaces and a fluid electrolyte. Therefore, electricity is stored in the form of electric charge between the two electrode plates.

The two main features are the extremely high capacitance values and the possibility of very fast charges and discharges due to extraordinarily low inner resistance. Still other advantages are durability, high reliability, no maintenance, long lifetime and operation over a wide temperature range and in diverse environments. And they don’t have the heat and expansion problems caused by fast charge/discharge cycles on lithium-ion. The discharge curve is another disadvantage. Whereas the electrochemical battery delivers a steady voltage in the usable power band, the voltage of the supercapacitor decreases on a linear scale from full to zero voltage. This reduces the usable power spectrum and much of the stored energy is left behind.

A recent example is the La Gomera Project, in island of La Palma where a diesel power

62 | Page station is equipped with 6 supercapacitors each 55.55 F and 1080 V DC providing peak capacity of 4 MW up to 5 seconds (20 MWs) to minimize load shedding situations.

5.4 Synthesis

In this chapter, firstly, it was investigated that what’s the configuration for EES device considering both efficiency and SOC. Secondly, important design requirements for the EES are also defined and characteristic data of various EES technologies are compared and evaluated against the design criteria. It can be concluded that in terms of technical requirements, LTO battery, FES and SC can be alternatives for EES device installed on Thialf. See Table 5-11, actually, FES and SC are better than LTO because of extremely good current handling capabilities and life cycles (refer to Table 5-4). But the disadvantages of FES and SC are also obvious which are dimension (refer to 5.3.3) and cost (refer to Table 7-2). LTO is much superior to FES and SC on these two points. Therefore, LTO battery is selected to be the alternative with highest priority for hybridization of Thialf. Last but not the least, technical characteristics are also viewed through the state of development overview of alternative technologies and several applications in the industry are also demonstrated.

Table 5-11 LTO, FES and SC comparison32 Technology Performance Dimension Cost LTO ++ +++ +++ FES +++ + + SC +++ + + 6. Modelling

An onboard power system model is constructed based on proposed system configuration in Matlab/Simulink, with the intention to verify and improve previous simple calculations. What’s more, different power control algorithms for the hybrid system can also be tested to compare system-level performance.

6.1 Goal of modelling

Goals of the modelling part are listed below:

1. The overall simulation should be capable of analyzing power-flows and energy- flows based on given operational profile and timespan. 2. The overall simulation should be able to evaluate different system configurations

32 Rating: + to +++, poorest to best.

63 | Page

(benchmark mode and hybrid mode with different EES technologies). 3. The overall simulation should be able to investigate the influence of different control strategies hybrid on system performance.

Tasks of the modelling part are also listed below according to the goals:

1. Verify the saving potentials in the hybrid system: a. Quantify fuel savings by the improvement of operational point; b. Quantify fuel savings and reduction of maintenance by more smooth DE operation; c. Quantify reduction of maintenance by less running hours of DE. 2. Verify the configuration of EES device by taking EES efficiency losses into consideration: a. Verify the energy capacity; b. Verify the power capacity. 3. Test different control strategies in hybrid system.

6.2 System integration and overview

A diagram of the overall modelling scope is shown in Figure 6-1. Simulation starts from the load measurements of generators which are active power. Considering its efficiency as shown in Appendix 0 Figure 0-10, the brake power of the diesel engine can be obtained. Power flows into the control strategies module which determines the power provided by diesel engine and Li-ion battery (LTO) to realize its predefined function in hybrid system. The SOC is feedback as a control signal for the control strategies module so that the status of battery can be monitored. Finally, results like fuel consumption, fuel savings, battery capacity and cycle life can be compared as the output of the model by running different simulations.

Power Control Load Load Diesel Engine [kW] strategies [%]

Power Results [kW] Current Converter [A] EES

SOC

Figure 6-1 Overall modelling scope

6.3 Diesel engine model

As fuel consumption is the focus in this work, modelling detailed combustion process of

64 | Page the diesel engine is initially thought not to be necessary. A very simple data-oriented DE model seems to meet the criteria as it normally provides adequate accuracy and correlation between inputs and outputs with low computation effort. This is the original idea of developing the Excel model to calculate the fuel saving potentials. The total consumed fuel according to operational profile was calculated in Excel model by:

= = ( ) Equation 6-1 𝑡𝑡2 𝑡𝑡2 𝑚𝑚Where𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 ∫𝑡𝑡1 𝑚𝑚 ̇ 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓is the𝑑𝑑𝑡𝑡 fuel∫ 𝑡𝑡mass1 𝜂𝜂𝑠𝑠𝑠𝑠𝑠𝑠 flow𝑊𝑊̇ 𝐸𝐸 𝑑𝑑rate𝑡𝑡 and is the engine SFC. Only treated as a steady-state or quasi steady-state approximation, the SFC at any non-idle condition can 𝑚𝑚̇ 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝜂𝜂𝑠𝑠𝑠𝑠𝑠𝑠 be determined at each engine load point from the engine SFC map. However, fuel consumption results from Excel model is about 12% lower compared with measurements onboard. Therefore, Excel model is not sufficient enough for this project mainly considering the reason below (see chapter 4.3 reason c33): • Quantify fuel savings (and reduction of maintenance) by more smooth DE operation cannot be achieved. Because this fuel consumption model in Excel is a kind of data-oriented model which only the global relation between delivered power at steady state and the fuel consumption is considered through polynomial equations. No test data of how the step load influence fuel consumption is available inside Excel model. Neither diesel engine physical process nor dynamic aspects of diesel engine process are incorporated. Thus, this model is an extremely simple transparent box and one cannot explore physical correlations inside the model between inputs and outputs.

Consequently, more complicated diesel models like process-oriented model or mean value model (between the data-oriented and process-oriented model) should be required to accomplish the predefined tasks. Considering accessibility, the first priority is DE A model which is a mean value model based on a causal analysis of the working principles of diesel engine developed in the Marine Technology Department of TU Delft. Several versions have been developed for DE A model, among which ‘DE A4 static version’ and ‘DE A4 dynamic version’ are discussed here.

6.3.1 DE A4 static version The causality of the DE A model is fuel injection. Heat released from fuel is calculated. After considering losses from combustion, heat transfer, thermodynamic cycle and mechanical, the resulted engine power is determined [52]. The so called ‘static version’ doesn’t mean this version is time independent because time dependency is the nature of DE A model. It is because the ‘shaft dynamics’ module and ‘PID controller modelled governor’ module is eliminated in this version so that it can be used with other static models for the JOULES project of the European Union.

Since in the original model, a governor, which uses a PID controller to process the difference between the target speed or torque and the actual speed or torque, manages

33 Since reason c is the most important one relative to research objective and nothing can be done to verify reason a and b in this project.

65 | Page the amount of fuel to be injected by the fuel pump. Leaving out the governor in static version, a fitting between fuel injection rate and power output must be done to achieve the desired direction of simulation as shown in page 39 of [52].

A problem within the DE A4 model is observed after the input of engine parameters and running simulations, which is shown in Figure 6-2. It can be seen that in the figure, the DE A model outputs are fairly accurate and follow the SFC data in the upper half of the engine load but can only provide a rough estimate of the trend at low loads area. It leads to a lower fuel consumption result from DE A4 model compared with the Excel model, which makes the deviation from measurements even larger. For this reason, it is suspected that DE A4 model calculates an increased engine efficiency at low load area.

SFC comparison (test vs DE A4) 380 360 340 320 300 280

[g/kWh] 260 240 220 200 180 0 20 40 60 80 100 [%MCR]

sulzer(DE_A4) sulzer(new data test) mak(DE_A4) mak(new test data)

Figure 6-2 SFC comparison

The total efficiency of DE is [2]:

= = 𝜂𝜂𝑒𝑒 Equation 6-2 3600000 𝑒𝑒 𝐿𝐿 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑞𝑞 𝑡𝑡𝑡𝑡 𝑚𝑚 Not𝜂𝜂 all 𝑠𝑠𝑠𝑠𝑠𝑠input∙ℎ heat𝜂𝜂 goes∙ 𝜂𝜂 into∙ 𝜂𝜂 the∙ 𝜂𝜂thermodynamic engine cycle: see Figure 6-3. Firstly, combustion losses is due to combustion may be incomplete. Secondly, a significant amount of the combustion heat is carried away through the cylinder liner and the cylinder head into the jacket cooling water and through the piston crown to the cooling oil, which is the heat input efficiency . Thirdly, the thermodynamic efficiency is completely determined by the shape of the cycle in the thermodynamic diagrams. At last mechanical 𝜂𝜂𝑞𝑞 𝜂𝜂𝑡𝑡𝑑𝑑 losses like friction is accounted by mechanical efficiency [2].

𝜂𝜂𝑚𝑚

66 | Page

Figure 6-3 Energy losses in a diesel engine (Source:[2])

Betz and Woschni found a complex relation between the air excess ratio and the combustion efficiency, which is used in the DE A4 model. Heat input efficiency at different operational points is derived from SFC data, after which using polynomial fitting, a curve of heat input efficiency is achieved. Mechanical losses is calculated with a constant factor and an engine speed dependent factor in the model. Since the engine operates at constant speed, absolute mechanical losses in this situation always keeps constant as the same in the nominal point of DE.

Investigating the accuracy of each partial efficiency at low loads can be time consuming and it is also not the core objective of the thesis. But a parametric method can be used to match the total efficiency of the model with test data (refer to Equation 6-2). Methods of calculating heat input efficiency, thermodynamic efficiency and mechanical efficiency are kept as the same in the original DE A4 model. And since the total efficiency can be obtained from the SFC test parameters thus combustion efficiency can be fitted so that total efficiency calculated in the DE A4 model can match the test data. The result is shown in Figure 6-4.

Effeiency (Sulzer6ZAL40S) total efficiency combustion efficiency heat input efficiency thermodynamic efficiency mechanical efficiency 100

60 Efficiency [%] 50

20 0.5 1 1.5 2 2.5 3 3.5 4 4.5

6 Output Power [W] 10 Figure 6-4 Efficiencies of Sulzer6 engine

After fitting the combustion efficiency of DE A4 model, it gives same SFC data as the test

67 | Page data in Figure 6-2. Details on fuel consumption and saving potentials of Excel model and DE A4 static version are outlined in Appendix G. It can be seen that DE A4 static model verifies fuel consumption and fuel saving result obtained from Excel model. It also implies that effects of DE dynamic transient operations on fuel consumption are not captured by DE A4 static model as well. 6.3.2 DE A4 dynamic version Since both Excel model and DE A4 static model fail to take the influence of load variations on fuel consumptions, which might lead to deviations of simulation results from measurements especially for turbo-charging diesel engines in this project. Thereby, steady-state fuel consumption computations which is carried out before, at best, can be considered as a baseline for further transient investigations. Because real transient operation is far from steady in this project, DE A4 dynamic model is then investigated.

The dynamic version also includes a shaft dynamics module and a PID controlled governor module. Shaft dynamics model uses torque required from the consumer and engine as inputs. The engine rotational speed is calculated according to Newton’s second law of rotational motion in the follow equation.

= + Equation 6-3 𝑡𝑡𝑒𝑒𝑒𝑒𝑒𝑒 1 𝑀𝑀𝑒𝑒𝑒𝑒𝑒𝑒−𝑀𝑀𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 Where𝑛𝑛𝑒𝑒𝑒𝑒𝑒𝑒 ∫𝑡𝑡𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 is2 the𝜋𝜋 ∙ total𝐼𝐼𝑡𝑡𝑡𝑡 𝑡𝑡inertia∙ 𝑑𝑑 of𝑡𝑡 the𝑛𝑛0 shaft and load and is the initial shaft rotational speed. The input of the fuel pump and governor model is the speed setting and actual 𝐼𝐼𝑡𝑡𝑡𝑡𝑡𝑡 𝑛𝑛0 speed of the engine. The speed difference will be processed by a PID governor and restricted by a limiter. The result is the signal output fuel rack position ( ) to the fuel pump. Based on and the amount of fuel should be injected at the nominal point 𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠 ( ), the fuel pump model calculates the amount of fuel injected per cylinder per cycle 𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠 through the following equation: 𝑋𝑋𝑛𝑛𝑜𝑜𝑚𝑚 𝑓𝑓 𝑚𝑚 = _ Equation 6-4 𝑋𝑋𝑠𝑠𝑠𝑠𝑠𝑠 𝑓𝑓 𝑛𝑛𝑛𝑛𝑛𝑛 𝑓𝑓 𝑛𝑛𝑛𝑛𝑛𝑛 Where𝑚𝑚 𝑋𝑋 ∙_ 𝑚𝑚 is the fuel injected into the cylinder per cycle per cylinder at nominal point, which is calculated through nominal parameters. Figure 6-5 below shows the shaft 𝑚𝑚𝑓𝑓 𝑛𝑛𝑛𝑛𝑛𝑛 dynamics at step loads. Three kinds of step loads are designed 10%, 20% and 30% all in one second, which also forms a 60% step in three seconds. As can be seen in the figure a 1.4% RPM deviation is observed from the nominal RPM at step load.

68 | Page

Shaft dynamics at step loads

100

Load [%] 60

40 15 20 25 30 525

520

515 RPM

510

505 15 20 25 30 Time [Second]

Figure 6-5 Shaft dynamics

Firstly, two simulations are carried out using static and dynamic version of DE A4 model respectively to compare the fuel consumption result for NO.4 MAK engine according to day1 recorded operational profile. And the two simulation results are almost the same, more accurately, the dynamic version is about 0.53% higher than the static version, which implies two potential directions of following research:

1. Although the dynamic DE A4 model provides more dynamics than the static version like the shaft dynamics and governor dynamics but it is still not good enough for quantifying the effect of transient dynamics on fuel consumption. Because the core part which is the diesel engine process module inside the dynamic version is exactly the same with the static version. And the nature of the diesel engine process model is still a mean value model which describes the engine torque as a mean value over one or more engine cycles rather than modelling them cylinder by cylinder. Because the cyclic variation of states in the engine is not of interest. 2. The result of dynamic version is right which means transient load variations has only marginal effect (less than 1%) on fuel consumption of DE. Since as described in 4.1.1 already, the load variations of DE during recorded period are fairly “acceptable”. And the 12% deviation is caused by the not accurate engine trail data, not accurate measurements and etc.

Next step is to verify two reasoning directions proposed above. Two operational profiles are designed for DE to compare the fuel consumption at steady load and step load as shown in Figure 6-6. Operational load1 is a dynamic repeating load from 100% to 50% and then come back to 100% again in a period of 30 seconds, which captures a much more dynamic load profile than recorded days. Operational load2 is just a steady load at 75% all the time. The simulation time span is set to one hour for the MAK engine on each profile. See Figure 6-7, it is interesting that the outputs of average fuel mass rate from DE A4 model in step load profile1 is almost the same as real time fuel mass rate in steady load profile2. Therefore, fuel consumption results from two operational profiles are almost the same, or more accurately, the fuel consumption at repeating step loads is 0.25% higher than the steady state loads. Considering the maximum recorded ramp rate is

69 | Page about 10% of DE engine capacity per second. Then the repeating step load period is reduced to 10 seconds, which means the step load is increased to 10% of the nominal capacity per second during the simulation. Again, the fuel consumption result is almost the same as the previous simulations, more accurately, it is 0.33% higher than the steady load profile simulation.

DE operational profile

100 Step load1

80 Steady load2 Load [%]

70 0 10 20 30 40 50 60 70 Time [sec]

Figure 6-6 Different operational profile

DE fuel mass rate 0.28 real time fuel mass rate (dynamic load) 0.26 running mean fuel mass rate (dynamic load) real time fuel mass rate (steady load) 0.24

0.22

0.2

0.18

Fuel mass rate [kg/s] rate mass Fuel 0.16

0.14

0.12 0 100 200 300 400 500 600 Time [Second] Figure 6-7 DE fuel mass rate comparison in dynamic and steady operational profile

Conclusions can be made on the two assumptions according to above simulation results: 1. DE A4 dynamic version model is also not the right toolbox for estimating fuel consumption in dynamic load. 2. The effect of dynamic load on DE fuel consumption during recorded days is still not clear, which means whether smooth operation of DE in hybrid system can lead to extra fuel savings is not clear.

In terms of conclusion2, studies found that turbocharged diesel engines (which is the case onboard of Thialf) cannot respond to a sudden change in speed or load as fast as naturally aspirated diesel engines because the change in compressor air flow lags behind the change in fueling rate. The transient acceleration or deceleration process of naturally aspirated diesel engines can be approximated by a continuous series of steady-state

70 | Page operating conditions. However, in turbocharged diesel engines, during transients the turbine power is not equal to the compressor power, and the turbocharger speed is affected by turbocharger inertia and turbocharger power imbalance. There is a turbocharger lag during which the compressor boost pressure gradually changes to reach a new steady state (The EGR circuit has a certain volume and there is a transient dynamic response of an EGR purging and filling process in the intake manifold). The transient response is usually in the order of several engine cycles. During transients, air- fuel ratio, EGR rate and in-cylinder metal wall temperature (due to thermal inertia) are all different from those in steady-state. The resulting deteriorated combustion efficiency and pumping loss cause differences in emissions and fuel economy between steady state and transient [21, 22]. Thus, it is noted that under the same fueling rate the transient engine torque is usually lower than the steady-state engine torque due to losses in combustion, pumping losses, turbocharger lag, and thermal inertia. In some cases the transient power can be 5-8% lower than the steady state power [21]. Therefore, the real transient SFC must be computed by a full process oriented, high-fidelity crank-angle- resolution engine cycle simulation model that reflects the transient combustion efficiency and heat losses, as well as the transient pumping loss during turbo lag which is related to engine controls.

6.4 Converter model

The semiconducting devices, when used to process high power (as opposed to signals), are called power electronics devices. In such applications, they are generally used only in two operating states, either on with near-zero resistance or off with a very large resistance across the main power terminals. The basic components of the converters are diodes, transistors and thyristors. A diode is a two-terminal device which allows a flow of current in one direction only; from anode to cathode. The switchover from on to off state is automatic and is not controllable by any signal. Other devices uses the diode as the building block, in which a third terminal is added to control (trigger) the on and off state by injecting a control signal at the control terminal. Thus, they virtually work as variable resistance that can be controlled ideally between zero (on) and infinity (off) by a small control signal applied at the third terminal. For this reason, the power electronics device in the electrical circuit diagram is often shown by a simple gate-controlled switch symbol between two power terminals and the third control terminal as shown in Figure. With the control signal on and off periodically, the device conduction can be turned on and off at high frequency (in the desire direction and at the desired time) as shown in Figure 6-8 [2, 53].

71 | Page

Figure 6-8 Control of power electronic device (Source:)

In this work, firstly a bidirectional buck-boost DC/DC converter model was used, which is from the matlab demo “Supercapacitor Model” provided by SimPowerSystems tool box. The model is built with the basic semiconducting modules like MOSFET (metal-oxide semiconductor field-effect transistor). However, in order to keep the acceptable accuracy the simulation speed is very low since the simulation step have to be kept at around 25 micro seconds. Thus this model is more suitable for investigating the dynamics of the converter itself.

Recalling the fact that the point of converter model in this work is to transfer power to current, therefore, a simple controlled current source module is used here to realize this function.

6.5 Battery model

The simulation of complete power systems, as with the hybrid vessel, doesn’t require a high level of precision. It is more important to know the general behavior of a battery, for example, it is important to represent the fact that the voltage available depends on the SOC and the current as shown in Figure 6-9 (left). In the idealized case (curve 1), the discharge of the battery proceeds at the theoretical voltage until the capacity is fully utilized, the voltage then drops to zero. Under actual conditions, the discharge curve is similar to the other curves. Because as the current drain of the battery increased, the IR losses and polarization effects increase, the discharge is at a lower voltage. At extremely low current drains (curve 2) the discharge can approach the theoretical voltage and theoretical capacity. With increasing current drain (curves 3-5), the discharge voltage decreases and the discharge shows a more sloping profile, and the delivered ampere- hour capacity, as well as the service life are reduced [27].

If a battery has reached the cut-off voltage34 under a given discharge current is used at a lower discharge rate, its voltage will rise and additional capacity can normally be obtained until the cut-off voltage is reached at the lighter load. As shown in Figure 6-9 (right), the

34 Cut-off voltage is the minimum allowable battery voltage. This voltage represents the end of the discharge characteristics. At the cut-off voltage, the battery is fully discharged (@ nominal discharge current). And it is a good practice to stop discharge at this point since further discharge can lead to chemical instability in the cell. The result being a reduced battery lifetime.

72 | Page discharge is first run at the highest discharge rate to cut-off voltage. The discharge rate is then reduced to the next lower rate. The voltage increases and the discharge is continued again to the cut-off voltage, and so on.

1 2 4 3 3 oltage oltage

V 4 V 2 End 5 1 voltage

0 Elapsed time of discharge 0 Elapsed time of discharge

Figure 6-9 Characters discharge curve (Source:[27])

Therefore, the battery model in this work is from SimPowerSystems toolbox in Simulink which provides a generic dynamic model can be parameterized to represent most popular types of rechargeable batteries. The main feature of this battery model is that the parameters required for modelling and simulation can be easily deduced from a manufacturer’s discharge curve.

6.5.1 Principle The battery is modelled using a simple controlled voltage source in series with a constant resistance. The equivalent circuit of the battery is shown in Figure 6-10. Where the “controlled voltage source” is the block converts the Simulink input signal ( ) into an equivalent voltage source. 𝐸𝐸𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏

Figure 6-10 equivalent circuit of the battery (Source:[54])

For Li-ion battery model, the battery voltage is calculated with described non-linear equations based on the actual SOC and current of the battery below:

73 | Page

Discharge mode ( > 0): ∗ = ( , , ) = 𝑖𝑖 + exp ( ) Equation 6-5 ∗ 𝑄𝑄 ∗ 𝑄𝑄 1 0 Charge𝐸𝐸 𝑓𝑓 𝑖𝑖𝑖𝑖 mode𝑖𝑖 𝑖𝑖 ( <𝐸𝐸 0−): 𝐾𝐾 ∙ 𝑄𝑄−𝑖𝑖𝑖𝑖 ∙ 𝑖𝑖 − 𝐾𝐾 ∙ 𝑄𝑄−𝑖𝑖𝑖𝑖 ∙ 𝑖𝑖𝑖𝑖 𝐴𝐴 ∙ −𝐵𝐵 ∙ 𝑖𝑖𝑖𝑖 ∗ = ( , , ) = + exp ( ) Equation 6-6 𝑖𝑖 . ∗ 𝑄𝑄 ∗ 𝑄𝑄 2 0 And,𝐸𝐸 𝑓𝑓 𝑖𝑖𝑖𝑖 𝑖𝑖 𝑖𝑖 𝐸𝐸 − 𝐾𝐾 ∙ 𝑖𝑖𝑖𝑖+0 1∙𝑄𝑄 ∙ 𝑖𝑖 − 𝐾𝐾 ∙ 𝑄𝑄−𝑖𝑖𝑖𝑖 ∙ 𝑖𝑖𝑖𝑖 𝐴𝐴 ∙ −𝐵𝐵 ∙ 𝑖𝑖𝑖𝑖 = ( ) = ( ) = 𝐸𝐸𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑉𝑉 = ( ) 𝑉𝑉𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑉𝑉 𝐸𝐸𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 − 𝑅𝑅 ∙ 𝑖𝑖 Exp(s) = ( ) 𝐸𝐸0 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑉𝑉 ( ) = . 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑉𝑉 ( ) = 0 , ( ) = 1 𝑆𝑆𝑆𝑆𝑆𝑆 𝑠𝑠 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝑡𝑡ℎ𝑒𝑒 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑚𝑚𝑚𝑚𝑑𝑑𝑑𝑑 K = ( ) ( ) 𝑆𝑆𝑆𝑆𝑆𝑆 𝑠𝑠 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑆𝑆𝑆𝑆𝑆𝑆 𝑠𝑠 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑐𝑐ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 = −1 ( ) 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐴𝐴ℎ 𝑜𝑜𝑜𝑜 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑂𝑂ℎ𝑚𝑚𝑚𝑚 i ∗= ( ) 𝑖𝑖 𝐿𝐿𝐿𝐿𝐿𝐿 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝐴𝐴 it = ( ) 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐴𝐴 Q = ( ) 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐴𝐴ℎ A = ( ) 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐴𝐴ℎ B = ( ) 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣𝑣 𝑉𝑉 The battery SOC is calculated as:−1 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝐴𝐴ℎ SOC = 100(1 ) ∫ 𝑖𝑖 𝑑𝑑𝑑𝑑 It should be noted− as well that the model provided by SimPowerSystems and used in this 𝑄𝑄 work is based on some assumptions and has limitations:

1. The internal resistance is supposed constant during the charge and discharge cycles and doesn’t vary with the amplitude of the current.35 2. The mode’s parameters are deduced from the discharge characteristics and assumed to be the same for charging. 3. The capacity of the battery doesn’t change with the amplitude of the current (No Peukert effect). 4. The self-discharge of the battery is not represented 5. The battery has no memory effect. 6. The temperature effect which affect the battery’s behavior is included in the model but not used in the simulations.

6.5.2 (Dis) charge characteristics The particular battery type can be accurately modelled by well determining parameters of the equivalent circuit based on its discharge characteristics which can be presented by a typical discharge curve composed of three sections as shown in Figure 6-11. For this battery model in the Matlab-Simulink SimPowerSystems library, a user-friendly interface

35 Real cell resistance increases due to the accumulation of discharge products, activation and concentration, polarization, and related factors.

74 | Page allows the user to enter standardized parameters which can be obtained from manufacture’s data sheet and discharge curve, and then the model’s required parameters are calculated automatically according to a similar method to that presented in [55].

Figure 6-11 shows the discharge characteristic of a 2.3V and 2.9Ah Li-ion cell. The first section represents the exponential voltage drop when the battery is discharged. Depending on the battery type, this area is more or less wide. In this example, the exponential voltage and capacity of the Li-ion battery is 2.48V and 0.14Ah. The second section represent the charge that can be extracted from the battery until the voltage drops below the battery nominal voltage. The capacity at nominal voltage is 2.6 Ah. Finally, the third section represents the total discharge of the battery, when the voltage drops rapidly. This shows the maximum capacity of the battery is 2.9 Ah. What’s more, discharge curves at different rates are also shown in Figure 6-11. It can be seen that higher discharge rate combined with internal resistance and polarization effect results in less available capacity until cut-off voltage.

Nominal Current Discharge Characteristic at 0.43478C (1.2609A)

2.8 Discharge curve 2.6 Nominal area 2.4 Exponential area 2.2

Voltage 2 End of exponential zone End of nominal zone 1.8

0 0.5 1 1.5 2 2.5 3 3.5 4

0.2C, 1C, 5C and 9C discharge curves

2.8 0.2C 2.6 1C 2.4 5C 2.2 9C

Voltage 2 1.8

0 0.5 1 1.5 2 2.5 3 3.5 4 Ampere-hour (Ah) Figure 6-11 discharge characteristics (refer to Figure 6-9 left)

Figure 6-12 shows the 2.3V, 2.9Ah cell discharges at 59A until a particular voltage and continues to discharge at 5.9A until the full ampere-hour capacity is delivered, which verifies the right part of Figure 6-9. The horizontal axis is time and unit is second. It also proves that the (maximum) capacity of the modelled battery doesn’t change with the amplitude of the current (No Peukert effect) as mentioned in 6.5.1.

75 | Page

Current Voltage 60 2.5

40 2 [V] [A] 20 1.5

0 1 0 200 400 600 800 1000 0 200 400 600 800 1000

SOC Extracted capacity 100 3

50 [%] [Ah] 1

0 0 0 200 400 600 800 1000 0 200 400 600 800 1000 Figure 6-12 Two-stage discharge (refer to Figure 6-9 right)

6.5.3 Extracting model parameters Some parameters can be directly extracted from battery manufacture data sheet like rated capacity and internal impedance. The other detailed parameters can be deduced from the typical discharge characteristics plot. Definitely, the parameters are approximate and depend on the precision of the points obtained from the discharge curve. A tool called Scanlt provided by amsterCHEM can be used to extract values from data sheet curves. A list of High power Li-ion battery is listed in Appendix 0 Table 0-7.

Neglecting small differences in single cells characteristics, we can assume the battery pack with all of its components and effects as a very big single cell. Thus, to model a series and parallel combination of cells based on the parameters of a single cell, the simple method of parameter transformation can be used in the table. This configuration does not allow battery modules to have different SOC and thus cannot be used for battery bank module balancing simulations.

Table 6-1 Parameter transformation Parameter Value Nominal voltage Single cell * Nb_ser Rated capacity Single cell * Nb_par

Maximum capacity Single cell * Nb_par Fully charged voltage Single cell * Nb_ser Nominal discharge current Single cell * Nb_par Internal resistance Single cell * Nb_ser/Nb_par Capacity @ nominal voltage Single cell * Nb_par Exponential zone Single cell *[Nb_ser, Nb_par]

A 2.3V, 2.9Ah Toshiba-SCiB cell is modelled here in this work. 400 cells in series36 and 129 cells in parallel constitutes the big cell which has the capacity of 344kWh as

36 DC bus voltage is set to be 920 V which refers to information obtained from system integrator.

76 | Page determined in chapter 5.1.

6.5.4 Battery efficiency & cut-off voltage There are two potential issues considering previous method used to calculate battery capacity:

1. A constant nominal 95% roundtrip efficiency is used in 5.1.1; 2. Cut-off voltage is not considered.

In terms of typical peak shaving, these two issues do not exist. Since (dis)charge current is below 0.3C (refer to 5.1.3) most of the time, which means roundtrip efficiency can be as high as nominal efficiency or even higher. What’s more, SOC is between 74.7% and 100% (refer to 5.1.2), which means cut-off voltage cannot be achieved. However, in terms of spinning reserve function and demanding peak shaving function, these two issues are real. As average discharge current is 9C, the roundtrip efficiency can be lower than the nominal efficiency. What’s more, cut-off voltage can be achieved since a majority of stored energy is required to deliver at high rates.

See Figure 6-13, it illustrates discharge and charge efficiency of the 344kWh battery at constant currents using Equation 6-7 and Equation 6-8:

= Equation 6-7

∫ 𝑉𝑉𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏∙𝑖𝑖𝑖𝑖𝑖𝑖 2 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝜂𝜂 ∫ 𝑉𝑉 ∙𝑖𝑖𝑖𝑖𝑖𝑖 + ∫ 𝑖𝑖 ∗𝑅𝑅𝑅𝑅𝑅𝑅 = Equation 6-8 2 ∫ 𝑉𝑉𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏∙𝑖𝑖𝑖𝑖𝑖𝑖− ∫ 𝑖𝑖 ∗𝑅𝑅𝑅𝑅𝑅𝑅 It𝜂𝜂𝑐𝑐 canℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 be obtained∫ 𝑉𝑉𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 from∙𝑖𝑖𝑖𝑖𝑖𝑖 the figure that the average roundtrip energy efficiency at 1C, 5C and 9C is respectively about 0.98, 0.90 and 0.8. Therefore, the capacity of 344kWh should be slightly modified to 366 kWh (See Equation 6-9 which is an update of Equation 5-4), which means 400 cells in series and 137 cells in parallel.

= = + = 366 ...... 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 78 230 Equation𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑐𝑐6𝑐𝑐𝑐𝑐𝑐𝑐-9 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 0 98∗0 97∗0 97∗√0 95 0 98∗0 97∗0 97∗√0 8 𝑘𝑘𝑘𝑘ℎ

(Dis)charge efficiency 1 1C discharge 1C charge 0.98 5C discharge 5C charge 9C discharge 0.96 9C charge

0.94

0.92 Efficiency [-]

0.9

0.88

920 900 880 860 840 820 800 780 760 740 720 700 Voltage [V]

Figure 6-13 (dis)charge efficiency

77 | Page

See Figure 6-14. At 0.2C and 1C discharge, the nominal 2.9*129 Ah capacity of the “big cell” can be fully offered until 690V cut-off voltage. However, about 70% and 90% nominal Ah capacity can be provided respectively at 9C and 5C discharge from fully charged (SOC=100%) to cut-off voltage. Delivered energy (Wh capacity) is equal to the integral of extracted Ah capacity times voltage.

Discharge curves

1100 0.2C

1000 1C 900 5C 800 Voltage [V] 700 9C

0(100) 50(87) 100(73) 150(60 ） 200(47) 250(34) 300(26) 350(7) Capacity [Ah] (SOC [%])

Figure 6-14 344kWh (Toshiba SCiB 2.3V, 2.9Ah cell) discharge characteristics

Therefore, in terms of the 366kWh battery, assuming an initial SOC of 90.9% (refer to 5.1.2) and simulating scenarios defined in 4.2.1 and 4.2.5 respectively, if required energy (refer to 4.2.1 230kWh and 4.2.5 206 kWh) can be delivered to the load until cut-off voltage then it can be considered that 366 kWh is enough.

Spinning reserve function Simulation results of 366 kWh for spinning reserve function are shown in Figure 6-15. The horizontal axis is time and unit is second. Although the required discharging period is 292 seconds (refer to 4.2.1), but cut-off voltage is achieved until 187th second thereby the simulation is stopped, see voltage curve. According to energy curve, during the discharge period until cut-off voltage, about 150 kWh energy is delivered to the load, whereas the required amount is 230 kWh. Thus, the capacity of the battery needs to be enlarged. Although from SOC curve, it can be concluded that about 40% of charge can still be available when the discharge current is reduced.

Voltage SOC 1000 100

900 80 [V] [%] 60 800

40 700 0 50 100 150 0 50 100 150

C-Rates Energy (load) 15 200

150 10

100 [-]

5 [kWh] 50

0 0 0 50 100 150 0 50 100 150 Figure 6-15 366 kWh battery for spinning reserve

Therefore, another 80 kWh is required according to Figure 6-15. The updated “big cell” is thereby 446 kWh which constitute of 400 cells in series and 167 cells in parallel. What’s more the initial SOC can be set at 93.7% with the increased capacity. Results are illustrated

78 | Page in Figure 6-16. It can be seen that during the 292 seconds discharge cut-off voltage is not achieved and required 230 kWh energy is delivered to the load. C-Rates is lower than the defined 9C discharge because of increased capacity. Thus, it can be concluded that 446 kWh is big enough for spinning reserve function.

Voltage SOC 1000 100

950 80 900 [V] 850 [%] 60

800 40 750 0 50 100 150 200 250 300 0 50 100 150 200 250 300

C-Rates Energy (load) 10 200

150 5 [-] 100 [kWh]

0 0 0 50 100 150 200 250 300 0 50 100 150 200 250 300 Figure 6-16 446 kWh battery for spinning reserve

Demanding peak shaving function In terms of demanding peak shaving function, the 446 kWh cell is tested at first to figure out whether it is good enough. Results are shown in Figure 6-17. It can be seen that during the 275 seconds discharging cut-off voltage is not achieved and required 206 kWh energy is delivered to the load. Thus, it can be concluded that 446 kWh is also big enough for demanding peak shaving function.

Voltage SOC 1000 100

80 900

60 [V] [%] 800 40

700 20 0 50 100 150 200 250 300 0 50 100 150 200 250 300

C-Rates Energy (load) 20 200

15 150

[-] 100 [kWh] 5 50

0 0 0 50 100 150 200 250 300 0 50 100 150 200 250 300

Figure 6-17 446 kWh battery for demanding peak shaving

Typical peak shaving It is not necessary to verify typical peak shaving function as mentioned above.

Synthesis Now the updated configuration and use pattern of battery system in each engine room can be determined. It is outlined in Table 6-2.

79 | Page

Table 6-2 Battery configuration Value Parameter Spinning reserve Typical peak shaving Demanding peak shaving Energy capacity 446 kWh Maximum C-Rates 10C 7C 16.8C

Maximum 7.8C N/A 7.8C continuous C-Rates Typical C-Rates 7.8C < 0.3C 7.8C Cycle life (full eq.) N/A 3.0 per day N/A SOC boundary N/A 100% to 80% N/A SOC setting N/A 93.7%37 N/A DOD 61% (from 93.7%) 20% (max) 59% (from 93.7%)

6.5.5 Battery lifecycle model The most accurate and reliable way to determine cycle life is to test several batteries from the same batch to determine fresh cell capacity. Then, the batteries are tested on a cycling machine that repeats a prescribed current trajectory that represents a typical cycle in the proposed application [56]. However, cycle life data given in the website of manufactures is usually quite simple and at most only taking into account how one or two factors independently influence the battery cycle life. Actually, battery life span and ageing algorithm is an associated complex, multi-factor process. Different factors like temperature, discharge and charge current, SOC and DOD jointly exert significant impacts on battery cycle life. But a full set of curves on how the 4 to 5 different factors independently and combined to influence the cycle life proves always to be confidential and not available from manufactures or system integrators. Thus, a replicated generic multi-factor battery cycle life prediction model according to Valentin [23] is used in this work to verify the cycle life performance of Li-ion batteries on Thialf’s application.

The main reason of using the model presented in [23] is that it is able to consider a multitude of dynamically changing cycling parameters. For Li-ion cells, the methodology has been tailored to consider five operational factors: charging and discharging current ( and ), minimum and maximum cycling limits ( and ), and operating temperatures (T) [23]. The effect on cycle life of each independent factor and different 𝐼𝐼𝑐𝑐ℎ 𝐼𝐼𝑑𝑑 𝑆𝑆𝑆𝑆𝑆𝑆𝑚𝑚𝑚𝑚𝑚𝑚 𝑆𝑆𝑆𝑆𝑆𝑆𝑚𝑚𝑚𝑚𝑚𝑚 joint factors are investigated in the model. What’s more, it is a generic model like the DE A4 model thereby at least only one point from cycle life curve is needed as an input of the model.

Methodology Firstly, an explanation of domain-specific terminology is provided in Table 6-3.

37 Can be modified.

80 | Page

Table 6-3 Domain-Specific Terminology (Source:[23]) Notation Meaning

CL Cycle life; the number of cycles under certain conditions until the battery reaches its end-of-life typically 70-80% of initial capacity.

nCL Normalized cycle life; the cycle life of a battery under particular conditions divided by the cycle life under nominal conditions. The minimum state-of-charge reached during a given battery charge and SOCmin discharge cycle. The maximum state-of-charge reached during a given battery charge and SOCmax discharge cycle. Average state-of-charge; the mean of the maximum and minimum state-of-charge SOCav of a cycle. Stat-of-charge; the amount of charge stored in a battery at a given point of time SOC divided by its nominal capacity. For the purpose of [23] it is assumed the battery voltage is fixed and the charge stored is directly proportional to energy stored.

DOD Depth-of-discharge; For one cycle: DOD = SOCmax – SOCmin.

The heart of the approach in [23]is the assumption that the impacts that certain environmental and operational factors on lithium-ion battery cycle life are, or can be approximated as being, independent from one-another [23]. Based on this independence assumption, it follows, where is the kth cycling parameter and degradation impact factor to be considered: 𝑥𝑥𝑘𝑘

( … ) ( ) ( ) , = × … × Equation 6-10 ( ,… ) ( ) ( ) 𝐶𝐶𝐶𝐶, 𝑥𝑥1 𝑥𝑥𝑘𝑘, 𝐶𝐶𝐶𝐶 𝑥𝑥, 1 𝐶𝐶𝐶𝐶 𝑥𝑥, 𝑘𝑘 𝐶𝐶𝐿𝐿 𝑥𝑥1 𝑛𝑛𝑛𝑛𝑛𝑛 𝑥𝑥𝑘𝑘 𝑛𝑛𝑛𝑛𝑛𝑛 𝐶𝐶𝐶𝐶 𝑥𝑥1 𝑛𝑛𝑛𝑛𝑛𝑛 𝐶𝐶𝐶𝐶 𝑥𝑥𝑘𝑘 𝑛𝑛𝑛𝑛𝑛𝑛 Thus:

( ) ( ) … = ( , … ) × × … × Equation 6-11 , , , ( ) ( ) 𝐶𝐶𝐶𝐶 𝑥𝑥, 1 𝐶𝐶𝐶𝐶 𝑥𝑥, 𝑘𝑘 𝐶𝐶𝐶𝐶�𝑥𝑥1 𝑥𝑥𝑘𝑘� 𝐶𝐶𝐶𝐶 𝑥𝑥1 𝑛𝑛𝑛𝑛𝑛𝑛 𝑥𝑥𝑘𝑘 𝑛𝑛𝑛𝑛𝑛𝑛 𝐶𝐶𝐶𝐶 𝑥𝑥1 𝑛𝑛𝑛𝑛𝑛𝑛 𝐶𝐶𝐶𝐶 𝑥𝑥𝑘𝑘 𝑛𝑛𝑛𝑛𝑛𝑛 In case of Li-ion ageing, among the most important impact factors are T, , , and DOD ( The state-of-charge boundaries and are transformed into 𝐼𝐼𝑐𝑐ℎ 𝐼𝐼𝑑𝑑 𝑆𝑆𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎 and DOD). Therefore, the combined Li-ion cycle life model is formed by taking the 𝑆𝑆𝑆𝑆𝑆𝑆𝑚𝑚𝑚𝑚𝑚𝑚 𝑆𝑆𝑆𝑆𝑆𝑆𝑚𝑚𝑚𝑚𝑚𝑚 produce of the normalized individual models (Equation 6-13, Equation 6-14, Equation 𝑆𝑆𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎 6-15 and Equation 6-16) and nominal cycle life (as specified by the battery data sheet of manufacture):

( , , , , ) = × nCL(T) × ( ) × ( ) × ( , ) Equation 6-12 𝐶𝐶𝐶𝐶 𝑇𝑇 𝐼𝐼𝑑𝑑 𝐼𝐼𝑐𝑐ℎ 𝑆𝑆𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎 𝐷𝐷𝐷𝐷𝐷𝐷 𝐶𝐶𝐶𝐶𝑛𝑛𝑛𝑛𝑛𝑛 𝑛𝑛𝑛𝑛𝑛𝑛 𝐼𝐼𝑑𝑑 𝑛𝑛𝑛𝑛𝑛𝑛 𝐼𝐼𝑐𝑐ℎ 𝑛𝑛𝑛𝑛𝑛𝑛 𝑆𝑆𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎 𝐷𝐷𝐷𝐷𝐷𝐷 Where nCL is each associated normalized cycle life fraction, corresponding to terms on the right side of Equation 6-10. And:

81 | Page nCL(T) = Equation 6-13 3 2 𝑎𝑎𝑇𝑇 −𝑏𝑏𝑇𝑇 +𝑐𝑐𝑐𝑐+𝑑𝑑 3 2 𝑎𝑎𝑇𝑇𝑛𝑛𝑛𝑛𝑛𝑛−𝑏𝑏𝑇𝑇𝑛𝑛𝑛𝑛𝑛𝑛+𝑐𝑐𝑇𝑇𝑛𝑛𝑛𝑛𝑛𝑛+𝑑𝑑 × ( ) × ( ) nCL( ) = Equation 6-14 × × ( ) 𝑒𝑒 exp, 𝑓𝑓𝐼𝐼𝑑𝑑 +𝑔𝑔 exp ℎ𝐼𝐼𝑑𝑑, 𝐼𝐼𝑑𝑑 𝑒𝑒 exp�𝑓𝑓𝐼𝐼𝑑𝑑 𝑛𝑛𝑛𝑛𝑛𝑛�+𝑔𝑔 exp ℎ𝐼𝐼𝑑𝑑 𝑛𝑛𝑛𝑛𝑛𝑛 × ( ) × ( ) nCL( ) = Equation 6-15 × × ( ) 𝑚𝑚 exp ,𝑛𝑛𝐼𝐼𝑐𝑐ℎ +𝑜𝑜 exp 𝑝𝑝𝐼𝐼𝑐𝑐ℎ, 𝐼𝐼𝑐𝑐ℎ 𝑚𝑚 exp�𝑛𝑛𝐼𝐼𝑐𝑐ℎ 𝑛𝑛𝑛𝑛𝑛𝑛�+𝑜𝑜 exp 𝑝𝑝𝐼𝐼𝑐𝑐ℎ 𝑛𝑛𝑛𝑛𝑛𝑛 ( , ) nCL( , DOD) = Equation 6-16 ( , ) 𝐶𝐶𝐶𝐶4 𝑆𝑆𝑆𝑆, 𝑆𝑆𝑎𝑎𝑎𝑎 DOD 𝑆𝑆𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎 𝐶𝐶𝐶𝐶4 𝑆𝑆𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎 𝑛𝑛𝑛𝑛𝑛𝑛 DOD𝑛𝑛𝑛𝑛𝑛𝑛 ( , DOD) = + ( + 100 ) 200 + + + 𝑢𝑢 2 𝐶𝐶𝐶𝐶4 𝑆𝑆𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎 + 𝑞𝑞 �2 ∗ 𝑣𝑣Equation∗ 𝑠𝑠 6-17∗ 𝑢𝑢 − ∗ 𝑡𝑡� ∗ 𝐷𝐷𝐷𝐷𝐷𝐷 𝑠𝑠 ∗ 𝑆𝑆𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎 𝑡𝑡 ∗ 𝐷𝐷𝐷𝐷𝐷𝐷 𝑢𝑢 ∗ 2 𝑎𝑎𝑎𝑎 𝑎𝑎𝑎𝑎 And𝐷𝐷𝐷𝐷𝐷𝐷 DOD∗ 𝑆𝑆𝑆𝑆𝑆𝑆 and average𝑣𝑣 ∗ 𝑆𝑆𝑆𝑆𝑆𝑆 SOC are characterized by the following four criteria:

0% 100%

𝑎𝑎𝑎𝑎 0% ≤ 𝑆𝑆𝑆𝑆𝑆𝑆 ≤100%

≤ 𝐷𝐷2𝐷𝐷𝐷𝐷 ≤

𝑎𝑎𝑎𝑎 𝐷𝐷𝐷𝐷𝐷𝐷 ≤ 2 ∗ (𝑆𝑆100𝑆𝑆𝑆𝑆 )

𝑎𝑎𝑎𝑎 𝐷𝐷Values𝐷𝐷𝐷𝐷 ≤ of coefficients∗ − 𝑆𝑆𝑆𝑆 of𝑆𝑆 the model are outlined in Table 6-4:

Table 6-4 Values for tuning coefficients of models of lithium-ion degradation impact factors (Source: [23]). Variabl a b c d e f g h m Value 0.0039 1.95 67.51 2070 4464 -0.1382 -1519 -0.4305 5963 Variabl n o p q s t u v Value -0.6531 321.4 0.03168 1471 214.3 0.6111 0.3369 -2.295

Figure 6-18 describes how normalized cycle life (y-axis) change against temperature (x- axis). It can be seen that for Li-ion batteries, the optimal temperature is around 20 degrees. Figure 6-19 and Figure 6-20 illustrates the impact of discharge current and charge current on cycle life respectively. It can be seen that cycle life is more sensitive to charge current. The impact of average SOC and DOD is shown by Figure 6-21 (y-axis is full equivalent cycles), of which the top view, side view and front view is respectively represented by Figure 6-22, Figure 6-23 and Figure 6-24. It is interesting to see that cycle life achieves the maximum value when average SOC is 50%, which means both very high and very low SOC is not good.

82 | Page

0.8

0.6 nCL(T)

0.4

0.2

0 -20 -10 0 10 20 30 40 50 T (degrees Celsius)

Figure 6-18 Normalized cycle life with temperature

0.8

0.6 nCL(Id)

0.4

0.2

0 0 1 2 3 4 5 6 7 8 9 10 Id(C-Rate)

Figure 6-19 Normalized cycle life with discharging current

1.4

1.2

0.8

0.6 nCL(Ich)

0.4

0.2

0 0 0.5 1 1.5 2 2.5 3 3.5 4 Ich(C-Rate)

Figure 6-20 Normalized cycle life with charging current

83 | Page

6000

4000

2000 CL (full eq.)

0 100 80 100 60 80 40 60 40 20 20 0 0 Average SOC (%) DOD (%)

Figure 6-21 Full equivalent cycle life with two-variable: average SOC and DOD

100

DOD (%) 40

0 100 90 80 70 60 50 40 30 20 10 0 Average SOC (%)

Figure 6-22 DOD-SOCav plane (top view of Figure 6-21)

7000

6000

5000

4000

3000 CL (full eq.)

2000

1000

0 0 10 20 30 40 50 60 70 80 90 100 DOD (%)

Figure 6-23 DOD-CL plane (side view of Figure 6-21)

84 | Page

7000

6000

5000

4000

3000 CL (full eq.)

2000

1000

0 0 10 20 30 40 50 60 70 80 90 100 Average SOC (%)

Figure 6-24 SOCav-CL plane (front view of Figure 6-21)

Case study and conclusion

Therefore, life cycles of a specific Li-ion battery used for Thialf hybrid system can be estimated. The specifications listed in Table 6-5 are obtained from website of TOSHIBA SCiB cells [57].

Table 6-5 Battery cell cycle life specifications (Source:[57]) Test conditions Value Temperature (degree C) 35 Discharge current (C) 10 Charge current (C) 10 Average SOC (%) 50 DOD(%) 60 Cycle life (Full eq.) >24,000

Regardless of the impact of temperature, estimated cycle life of SCiB LTO cells used for three defined functions (refer to Table 6-2) on Thialf can be summarized in Table 6-6. Given the required cycle life is 3 cycles per day and around 5800 for 10 years, LTO cycle life performance is far better than the required target.

Table 6-6 Cycle life estimation for Thialf application Function Cycle life (full eq.) Spinning reserve 41,649 Typical peak shaving 22,932 Demanding peak shaving 805,072

6.6 Control strategies

Three control strategies are proposed here. First one is called “constant SOC” strategy which is the initially designed EES taking load variations with more or less constant SOC and DE is allowed to ramp up/down at most in 10kW/s, which is shown in Figure 6-25 and Figure 6-26.

85 | Page

But after the analysis in 6.3 it shows that there exist the possibility that load variations of DE on fuel consumption is very limited. Thus the second control strategy can be proposed so that EES acts as a big UPS to replace one DE to secure enough redundancy, which is called “UPS” strategy. And load variations will be still absorbed by DE. Therefore, a smaller sized EES is required then for this situation, which reduce the EES system cost. What’s more, since the EES at most times will be at standby status the life cycles then will be never a problem. In this situation, the fuel saving potentials is almost the same with the first control strategy with a reduced cost.

On the other hand, as stated in 6.3 another opportunity is that fuel consumption for DE is still around 5-8% magnitude higher than in the steady state operation. And as can be seen in Figure 6-26 that the DE operates in hybrid mode still exposes to some degrees of dynamics. Thus control strategy 3 can be proposed so that the diesel engine can ramp up steadier as shown in Figure 6-28 and Figure 6-29. And a “varying SOC” strategy is used here as shown in Figure 6-27 which means during charge periods DE power can be increased and leads to a decrease in SFC as shown in Figure 6-30. But the fuel saving results indicates that the fuel saving potential is almost the same between “constant SOC” strategy and “varying SOC” strategy (around 4kg deviation using AFT engine room recorded data day1). This is because that the increase of DE power is very marginal during charge periods and also again the benefit of more steady DE operation on fuel consumption cannot be quantified. Therefore, “varying SOC” strategy only leads to more daily running cycles of EES which is a penalty actually. Furthermore, a potential risk of this strategy is that there might be not enough capacity for spinning reserve function and demanding peak shaving function. Additionally, the envisaged benefit of “varying SOC” strategy might be only realized when the capacity is big enough so that one more DE can be replaced but this is conflict with the scope of the project as mentioned in 2.6 and 4.3. Last but not least, the steadier DE operation can also be realized easily in “constant SOC” strategy, thus “varying SOC” strategy is eliminated on Thialf application.

To sum up, “constant SOC” strategy and “UPS” strategy are alternatives for Thialf application. Actually it depends on further research to determine which one is more beneficial for Thialf.

86 | Page

Control strategy1 80

steady SOC

70 SOC [%]

60 0 1 2 3 4 5 6 4 Time [Second] 10 Figure 6-25 SOC of EES control strategy1

DE power comparison 5000

2 DEs power benchmark mode 1 DE power hybrid mode (with EES control strategy1) 4000

3000 Power [kW]

2000

1000 0 1 2 3 4 5 6 4 Time [Second] 10 Figure 6-26 DE power comparison control strategy1

Control strategy2 80

varying SOC

50 SOC [%] 40

20 0 1 2 3 4 5 6 4 Time [Second] 10 Figure 6-27 SOC of EES control strategy2

87 | Page

DE Power comparison 5000 1 DE power hybrid mode 2 DEs power benchmark mode (with EES control strategy2)

4000

3000 Power [kW]

2000

1000 0 1 2 3 4 5 6 4 Time [Second] 10 Figure 6-28 DE power comparison control strategy2

DE Power comparison 2800 hybrid mode DE power (with EES control strategy2) 2600 hybrid mode DE power

2400 (with EES control strategy1)

2200

2000

1800 Power [kW]

1600

1400

1200 0 1 2 3 4 5 6 4 Time [Second] 10 Figure 6-29 DE power comparison of hybrid mode for control strategy 1 and 2

225 SFC comparison

220

215

210

205

SFC [g/kWh] 200

195

190

0 1 2 3 4 5 6 4 Time [Second] 10 Figure 6-30 SFC comparison of DE in hybrid mode for control strategy 1 and 2

88 | Page

6.7 Synthesis

In Chapter 6, the power system model is built in Simulink to verify and improve results obtained from previous chapters. Tasks (refer to 6.1) and corresponding results are summarized below. 1. Verify the saving potentials in the hybrid system: a. Quantify fuel savings by the improvement of operational point; • Modifications are made to the DE A4 static version so that fairly accurate SFC results can be achieved at low load areas. • Fuel saving results obtained from Excel model (14.46% and 13% for day1 and day2 respectively) is verified by improved DE A4 model which is 16.61% and 14.74% for day1 and day2 respectively. b. Quantify fuel savings and reduction of maintenance by more smooth DE operation; • It is verified that effect of dynamic transient DE operations on fuel consumption cannot be captured by DE A4 model. In other words, the fuel consumption output from DE A4 model is still 12.05% and 9.55% lower than the day1 and day2 measurements respectively. • The magnitude of fuel savings and reduction of maintenance by smooth DE operation thereby cannot be quantified and should be further investigated. c. Quantify reduction of maintenance by less running hours of DE. • Reduction of running hours is quantified (13928.4 hours yearly) and cost reduction of 23.3$ per less running hour is defined by cost engineer of HMC. 2. Verify the configuration of EES device by taking EES efficiency losses into consideration: d. Verify the energy capacity; e. Verify the power capacity; • In terms of sub-task d and e, the battery configuration is improved by considering cut-off voltage and energy efficiency at different (dis)charge rates. Finally a 446 kWh battery with maximum continuous 7.8C discharge is determined. • A generic Li-ion battery cycle life model is also used to better estimate the expected cycle life against combined multi-factors of Thialf application. A case study shows that more than 800,000 cycles can be achieved using SCiB LTO cells, which is far better than the 10 years’ cycle life target. 3. Test different control strategies in hybrid system. • Three kinds of control strategies are designed and compared. “Constant SOC” strategy and “UPS” strategy are alternatives for Thialf and “varying SOC” strategy is eliminated. But again it proves that further investigation on quantifying benefits of smooth DE operation is required since it also influences the selecting of “constant SOC” strategy or “UPS” strategy which only use battery as spinning reserve.

89 | Page

To sum up, the technical feasibility is successfully verified. 7. Economic analysis

Economic analysis contains two parts, benefits and cost (capital investment).

7.1 Benefits

Benefits of hybrid system is analyzed and quantified in previous chapters, which is outlined in Table 7-1.

Table 7-1 Benefits of hybrid system Yearly benefits Value Remark 1) Year fuel consumption is 22500 ton which is an Fuel savings (improved average value of last five years. 2) 15% fuel saving 894,375 $ efficiency of DE) in DP mode is used. 3) 500$/ton MDO price is used as an average value of last twelve months.

Fuel savings (smooth operation - - of DE)

Maintenance reduction (less 332,531 $ In total 13928 less running hours for DE in three running hours) engine rooms.

Maintenance reduction - - (smooth operation) The benefits of emission reductions cannot be quantified by dollars. It indeed reduces the Emission reduction N/A environmental footprint of HMC and may improve the competitiveness of the company.

Total 1,218,906 $ N/A

7.2 Cost

A battery installation is comprised of several components like battery cells, a power conversion system, materials in the module, a battery management system and other components. In addition, labor, maintenance and other variable costs must be taken into account. While individual cell costs may be a good economic indicator for comparison purposes, but they only represent around 20% (30% to 40% is also used in different source like [58]) of all relevant costs. But in some cases they can also represent around 40% to 50% of the total cost. Since total system and variable costs depend on location, application, additional equipment needed, vendors, commercial availability, size of the

90 | Page system and other variables [59].

There are a number of indicators used for cost comparison, such as cost per kWh (energy) and cost per kW (power). Table 7-2 represents current technology cost ranges. It can be seen that in terms of cost per kWh, FES and SC is much higher (more than an order of magnitude) than Li-ion batteries. In contrast, in terms of cost per kW, FES and SC is much lower. It should be noted that only a general cost range for “Li-ion” batteries from 2330 to 3000 $/kW is obtained from [60].

Table 7-2 Capital cost ranges of EES technologies

EES Type Cost (USD/kWh) Cost Prominent manufactures (USD/kW)

LFP/graphite 550-850[59] - A123 Systems, BYD, Amperex, Lishen

LMO/graphite 450-700[59] - LG Chem, AESC, Samsung SDI

LMO/LTO 900-2200[59] - ATL, Toshiba, Leclanché, Microvast

LCO/graphite 250-500[59] - Samsung SDI, BYD, LG Chem, Panasonic, ATL, Lishen

NCA/graphite 240-380[59] - Panasonic, Samsung SDI

NMC/graphite, 550-750[59] - Johnson Controls, Saft silicon 9900-10000[60] SC ~18000[61] 200-370[60] Maxwell typical 10000[32]

1300-3300[60] ABB, Beacon Power, Active FES 110-600[60] 7000-8000[61] Power, CCM

Investment cost of hybrid system is summarized in Table 7-3 below. It should be noted that since battery is about 20% to 50% of the total system cost. Thus an average value of 35% is used, which means other cost is about 65% and 1.85 times battery cost.

Table 7-3 Investment cost Investment cost Value Remark

691,300*3 = An average value of 1550 $/kWh is used for LTO Battery cost batteries. And the capacity is 446 kWh per 2,073,900 $ engine room.

Other relevant cost 3,836,715 $ Assumed an average 1.85 times battery cost.

Total 5,910,615 $ N/A

91 | Page

7.3 Synthesis

Benefits and costs is quantified in this chapter, which shows a payback period of 4.8 years. But it should be emphasized on the following points:

1. In terms of benefits: a. Some potential benefits are not included like extra fuel savings and further reduction of maintenance from smooth operation. b. Environmental benefits cannot be simply expressed by dollars. c. Oil price now is at a very low level, which means with the increase of oil price fuel savings can also be increased. 2. In terms of cost: a. The battery cost and other relevant cost is considered on a very high- level which is only for a roughly reference at concept design phase. b. While companies are also actively engaging in R&D to reduce the cost of implementing storage systems. The pace at which advancements are made and costs reduced varies from technology to technology[60]. Regarding “Li-ion” battery costs, battery prices have decreased significantly in the last decade. The yearly decrease is 14% between 2007 and 2014 [38], to about 300 USD/kWh in 2015. The driving force behind this development is mass production. Thus, cost of battery can be reduced further as it is a highly developing technology with a booming market.

c. As mentioned in different places in the thesis, the battery configuration is actually conservative like 5 minutes’ spinning reserve period, 466 kWh is not exactly the minimum capacity, etc., which means potential cost reduction also exists.

To sum up, considering above analysis the economic feasibility is fairly positive. 8. Conclusions & recommendations

This conclusions and recommendations chapter gives a summary and interpretation on the results of the research related to the research objective and sub-questions on the heavy lifting crane vessel Thialf of Heerema Marine Contractors. Also, in this section recommendations with respect to the assumptions used in the thesis will be given. Research methods and models will be reviewed and further research points are detailed.

92 | Page

8.1 Conclusions

At first, after reviewing EES technology development in the marine sector and defining the scope of the research problem of Thialf’s power system performance, the following research question were investigated:

The research structure was divided into 4 sub-questions.

• Answer to sub-question 1) “Functional decomposition and energy flow diagram.” (Refer to Chapter 3)

After modelling the energy transfer and distribution of Thialf in an energy flow diagram, the integration of EES technologies into onboard power system was illustrated in Figure 8-1. Three EES systems are connected to the three main switch boards in three engine rooms respectively to realize its two functions: spinning reserve and enhanced dynamic performance functions. Spinning reserve function was used to replace one running DE in each engine room and secure power supply during DE fault. Enhanced dynamic performance function (can be divided into typical peak shaving and demanding peak shaving sub-functions) is used to absorb power spikes and valleys and smooth DE operation.

G7 G5 G3 G1 G2 G4 G6 G8

MSB1-PS MSB1-SB

E/E E/E

CR CR EES G9 G10 G11 G12 EES

MSB2-PS MSB2-SB MSB3

E/E S1 P2 P3 S2 S3 P1 EES

Figure 8-1 Hybridization of benchmark system

• Answer to sub-question 2) “Operational profile of Thialf.” (Refer to Chapter 4)

Specifications of EES device and its design criteria for required three functions on Thialf was defined and assessed according to measured operational profile, see Table 8-1. EES was designed to replace one running diesel engine as spinning reserve in the hybrid system, which is the spinning reserve function (refer to Answer to sub-question 1)).

93 | Page

Typical peak shaving function defines that in typical dynamic positioning mode of Thialf, EES was used to absorb load dynamics (spikes and valleys) so that operation of diesel engine is smoothed (refer to Answer to sub-question 1)). Demanding peak shaving means EES is capable of handling load dynamics in the most extreme crane operations (refer to Answer to sub-question 1)). The energy capacity of EES thereby was estimated for the three functions according to the idea of how small the capacity could be instead of how big the system should be.

Efficiency improvement of power system through hybridization was also verified by estimating fuel saving potentials of the hybrid system which was 14.46% and 13% according to an on-board load measurement of 2 days which representative for the Thialf-operation. Fuel saving potentials was estimated by comparing fuel consumptions in benchmark power system and hybrid power system. Since with the specific fuel consumption (SFC) data of diesel engines, the load data were transformed into fuel consumption data within an Excel-model. However, the SFC-related fuel consumption outputs obtained from the Excel model were about 10% lower than accumulated measurements on board Thialf. Therefore, potential reasons lead to deviations between Excel model fuel consumption data and measurements were also analyzed, details refer to chapter 4.3.

Table 8-1 EES configuration defined in chapter 4 Parameter Spinning Typical peak Demanding reserve shaving peak shaving Energy capacity 230 kWh 78 kWh 206 kWh Maximum power 3680 kW 3079 kW 5380 kW

Maximum 2789 kW N/A 2750 kW continuous power Average power N/A 100 kW (max) N/A Cycle life (full eq.) N/A 3.9 N/A

• Answer to sub-question 3) “EES configuration and Ragone plot.” (Refer to Chapter 5)

The most relevant technical targets of EES device on Thialf were verified and improved based on chapter 4 considering both efficiency and SOC, see Table 8-2. Using Ragone plot and comparing the technical target of Thialf applications three EES-alternatives were identified: Lithium-ion batteries, flywheel energy storage and supercapacitors. Li-ion batteries were preferred because of prominent advantages on dimension (energy density) and cost, among which Lithium–Titanate battery technology was recommended because of good cycle life performance, symmetric discharge and charge capability and high levels of safety. Therefore, Lithium–Titanate battery model was used in the modelling part.

94 | Page

Table 8-2 EES configuration defined in chapter 5 Parameter Value Energy capacity 344 kWh Maximum C-Rates (discharge) 17C (for demanding peak shaving) Maximum continuous C-Rates 9C (for demanding peak shaving & (discharge) spinning reserve) Typical C-Rates (discharge) < 0.3C (for typical peaks shaving) Cycle life (full equivalent) 3.9 per/day (in DP mode) SOC boundary 100% to 74.7% SOC setting 90.9% (can be modified) DOD (maximum) 25.3%

• Answer to sub-question 4) “Modelling and cost analysis.” (Refer to Chapter 6 and Chapter 7)

In Chapter 6, the power system model was built in Simulink to verify and improve results obtained from Chapter 5, see Table 8-3, which was the final defined configuration for EES.

Table 8-3 Battery configuration defined in chapter 6 Value Parameter Spinning reserve Typical peak shaving Demanding peak shaving Energy capacity 446 kWh Maximum C-Rates 10C 7C 16.8C

Maximum 7.8C N/A 7.8C continuous C-Rates Typical C-Rates 7.8C < 0.3C 7.8C Cycle life (full eq.) N/A 3.0 per/day N/A SOC boundary N/A 100% to 80% N/A SOC setting N/A 93.7%38 N/A DOD 61% (from 93.7%) 20% (max) 59% (from 93.7%)

The application of this EES resulted in the following energy savings for the operation of Thialf by Heerema Marine Contractors:

• Fuel savings by eliminating DE spinning reserve requirement is: about 15% (16.61% and 14.74% for recorded two days respectively), which means an estimated yearly 900 k$ fuel saving considering 500$/ton marine diesel oil and 22500 ton average consumption per year.

• Reduction of running hours of diesel engines: about 14,000 hours yearly, which

38 Can be modified.

95 | Page

means a saving on maintenance of approx. 300 k$ per year.

• Furthermore, a fuel and maintenance saving potential by using peak shaving performance of the battery is showed, but this could not be verified in the model.

In terms of fuel consumptions and savings, modifications were made to the first principle mean value diesel engine model (the diesel engine A4 model of TU Delft) so that fairly accurate SFC results were achieved at low load areas. Therefore, fuel saving results obtained from Excel model (refer to 4.3) was also verified by improved DE A4 model. However, the magnitude of fuel savings and reduction of maintenance by smooth DE operation was not clear and should be further investigated. Because it was verified that effect of dynamic transient DE operations on fuel consumption cannot be captured by DE A4 model. Therefore the fuel consumption outputs from DE A4 model were still about 10% lower than the board measurements.

Three kinds of control strategies are designed and compared. “Constant SOC” strategy and “UPS” (uninterruptible power supplies) strategy were alternatives for Thialf and “varying SOC” strategy was eliminated. But again it proved that further investigation on quantifying benefits of smooth DE operation is required since it also influences the selecting of “constant SOC” strategy or “UPS” strategy which only use battery as spinning reserve. Therefore it can be concluded that the technical feasibility is successfully proved in chapter 6 on the basis of chapter 4 and 5.

In terms of economic analysis, a yearly benefit of about 1.2 million $ was achieved with capital investment of roughly 6 million $ which means a payback period of about 5 years. But it should be emphasized on the following points:

1. In terms of benefits: a. Some potential benefits were not included like extra fuel savings and further reduction of maintenance from smooth operation. b. Environmental benefits cannot be simply expressed by dollars. c. Oil price now is at a very low level, which means with the increase of oil price fuel savings can also be increased. 2. In terms of cost: a. The battery cost and other relevant cost was considered on a very high- level which was only for a roughly reference at concept design phase. b. While companies are also actively engaging in R&D to reduce the cost of implementing storage systems. The pace at which advancements are made and costs reduced varies from technology to technology[60]. Regarding “Li-ion” battery costs, battery prices have decreased significantly in the last decade. The yearly decrease is 14% between 2007 and 2014 [38], to about 300 USD/kWh in 2015. The driving force behind this development is mass production. Thus, cost of battery can be reduced further as it is a highly developing technology with a booming market. c. As mentioned in different places in the thesis, the battery configuration is

96 | Page

actually conservative like 5 minutes’ spinning reserve period, etc. thereby 466 kWh was not exactly the minimum capacity, which means potential cost reduction still exists.

Therefore, considering above analysis the economic feasibility is fairly positive.

Now the main objective of the thesis can be answered:

“Is it technical and economical feasible and advantageous of considering current electric energy storage technology for improving and optimizing the Thialf on board power generation and distribution system?”

In terms of technical feasibility, the answer is indeed yes. There are currently three kinds of available technologies suitable for Thialf application: Li-ion batteries (power version are favored like Lithium–Titanate battery), flywheel energy storage and supercapacitors, among which Lithium–Titanate battery has the highest priority. What’s more, with the rapid development of Li-ion technology, the hybrid system will also become more mature and optimized.

In terms of economic feasibility, it is not a yes or no answer. Because it depends on what the definition of feasibility is. By using Lithium–Titanate battery technology, the roughly estimated payback period is 4.8 years. This period can be shorter and the reasons are elaborated above. Therefore it can be conducted the economic feasibility of Thialf hybrid system is very positive.

8.2 Innovative work

To sum up, in addition to more or less applying existing knowledge and playing with models available from TU Delft and other sources, my own unique work are outlined below:

• Making concept design of the hybrid system and deriving design criteria; • Aligning a system integration approach to a cost saving potential based on the analysis of the actual Thialf data; • Integrating the power system model for Thialf which can be used for benchmark mode and hybrid mode; • Investigating the accuracy problem of DE A4 model in low load areas; • Investigating DE A4 model on dynamic transient operations.

8.3 Recommendations

1. More days’ recorded load consumptions are necessarily required to verify and improve results which derived from “typical” two days’ load consumption data. 2. During the modelling it is found that in terms of fuel consumption, DE A model outputs fairly accurate results in the upper half of the load curve and suffers from accuracy

97 | Page

problems during low load areas like below 50% (only gives a rough estimation of trend). Although usually the diesel engines are not designed to use at low load areas but for some applications like in this project, DE runs mostly under 25%. Thereby, it can be highly interested to improve the low load area performance of the model by improving the combustion process during low load areas. Although the author tried to improve it in this work and obtained good results. But the method which was used is a simple “fitting algorithm” and in principle “conflict” with the “first principle” nature of DE A4 model. This is an objective for future research. 3. DE A4 model in principle is a mean-value model which might fail to capture the impact of dynamic load in this project. Thus developing a high-fidelity crank-angle- resolution fully process oriented diesel engine model can be very interesting and very challenging as well. Assumption of SFC during transient dynamic load variations can be estimated using steady-state values is used in chapter 4.3. The process oriented diesel engine model can be also used to verify this assumption. 4. Battery model used in this work is without modelling Peukert effect, a more realistic battery model including Peukert effect can be further used. And effect of temperature on battery performance is also not considered in this work, thereby it can be further investigated. 5. Battery lifecycle prediction model used in this work is a static version which means all the parameters like discharge and charge current, average SOC and DOD are set to the same value in each cycle. For Thialf’s application, both discharge and charge are usually a variable current/time profile to simulate the real operational. Thus, a dynamic cycle life model can be investigated further. 6. The methodology used in this work to verify technical and economic feasibility can be improved. Because in the work the technical feasibility is verified at first based on a series of criteria, then the economic feasibility is verified against the most promising technical feasible alternative. However, technical feasibility is not an absolute yes or no answer, it is all about tradeoffs. Thus, basically each type of EES can be designed individually based on its favorable working conditions meanwhile in order to maximize the desired type of performance required by Thialf. Finally, the economic feasibility can be determined on each solution. 7. The initial idea of the author in the project in to develop a full power system model including the DC, AC grid and consumers (thruster and crane motors etc.) so that the full dynamic aspects of the power system can be observed. But finally, it proves too ambitious and the workload is huge. But developing such a model is of significant influence especially to concept design phase of projects by integrating/improving all the component models we have in the TU Delft maritime department library.

98 | Page

9. Discussions

9.1 Topic 1: Safety

Safety can be discussed on two considerations:

1. Whether the EES increases the safety/reliability of the power configuration of the Thialf vessel.

2. Safety problems of Li-ion batteries.

In terms of consideration one, EES slightly improves the safety level or reliability of Thialf power system. In DP mode, one of the main functions of EES is to provide spinning reserve function for 5 minutes in case of DE failure so that the reliability of hybrid system is as good as benchmark system which use DE as spinning reserve. In standby inshore and transient mode, there is no redundancy requirement for Thialf power system. Therefore, in other words, if a DE failure happens at the bench mark system, part of the load have to be removed until the standby engine is brought online. However, in the hybrid system, EES can still be used as spinning reserve function for these two modes so that required power can be delivered until standby engine is started and online again. Thus, EES improves the reliability of the power system in standby inshore and transient mode.

Additionally, on one hand, safety might be a serious issue in Li-ion battery technology because the charged positive and negative electrode materials react with the cell electrolyte at elevated temperature. On the other hand, the safety record of Li-ion batteries is extremely good and is improving39 due to the worldwide efforts of Li-ion battery manufacturers, OEM equipment designers, and regulatory agencies. If one assumes about 100 safety incidents per year from a global production of 4 billion cells, this translates to a safety incident rate of only 1 per 40 million cells, which is impressive [27, 40, 49].

9.2 Topic 2: Fuel cell

Fuel cell is not a kind of EES technology but a power source like diesel engine, therefore, it is not the target of this thesis. But newly designed HMC heavy lifting vessel Sleipnir may introduce LNG dual fuel engine on board, which offers an opportunity for the application of fuel cell. According to Viking Lady’s experience, in order to utilize hydrogen

39 Since power density is a specialized case for battery types that can be “tuned” with sublet chemistry changes and electrolyte modifications. Thus, Li-ion battery manufactures do this while balancing thermal stability and safety to reduce self-heating during high power (dis)charge.

99 | Page from reforming a fuel already on board such as LNG which is already powering the main engines, high temperature fuel cells such as molten carbonate fuel cell (MCFC, runs hot often around 600°C) and solid oxide fuel cell (SOFC) must be favored. Because its ability to handle the “dirty” hydrogen that comes with reforming process. This avoids the issue that dogs the more common lower-temperature proton exchange membrane (PME) cells, which means that PEM installations often need bulky hydrogen tanks to guarantee fuel purity [62]. Additionally, for PMEs, going from electricity to hydrogen and back to electricity can be horribly wasteful.

While fuel cells have varying characteristics, one thing the high-temperature versions usually don’t offer is a sharp energy release, generally slower to respond. Because sharp change in load, either up or down introduces thermal shock which damages the physical structure of high-temperature fuel cells. But one immature fuel technology—high- temperature PEM (HTPEM), inherit the load variation resilience of the cooler PEM cells and also possessing some of the fuel versatility of the much higher temperature systems, such as handling dirty hydrogen resulting from reforming processes, may be an option in the future [62].

Thus, if the fuel cell can be configured close to the average power draw—using batteries for low- or peak-load periods (such as initial or high-power bursts of activity) and using the cells to recharge the batteries. Using this way the fuel cells tend to have a longer life and the batteries are recharged with excess energy after the cells have ramped up. Obviously, it also needs huge amount of work to explore the feasibility of this opportunity in the future projects. Bibliography

1. Hebner, R., Energy Storage on Future Electric Ships. ESRDC report, March, 2008. 2. Stapersma, D., et al., Design of propulsion and electric power generation systems. 2002: London (United Kingdom): IMarEST. 3. Lindtjørn, J.O., et al., Demonstrating the Benefits of Advanced Power Systems and Energy Storage for DP Vessels. 2014. 4. TÖRNKVIST, A. THE LADY RECHARGES HER BATTERY. 2014; Available from: http://www.wartsila.com/docs/default-source/product-files/electrical-automation/electric- propulsion-systems/article-TW-2014-03-LLH.pdf?sfvrsn=8. 5. SUSTAINIA. SUSTAINIA 100 2013. 2013 [cited 2013; Available from: http://www.sustainia.me/solutions/. 6. Marine, I. Imtech Marine to supply hybrid propulsion system to the world's first seagoing low emissions hybrid ferries. Available from: http://imtech.com/EN/Marine/Imtech-Marine- Campaign/Green-Solutions/Imtech-Marine-to-supply-hybrid-propulsion-system.html. 7. FellowSHIP. FellowSHIP. Available from: http://vikinglady.no/. 8. SØRFONN, I. Wärtsilä's Hybrid Power System gives fuel savings and lower emissions. Available from: http://www.indetailmagazine.com/en/article/wartsilas-hybrid-power-system-

100 | Page

gives-fuel-savings-and-lower-emissions/2/2014/. 9. Marine, I. Hybrid Ferries CMAL Green vessels with innovative propulsion systems. Available from: http://imtech.com/Content/ImtechMarine/Documents/CMAL_Ferries_020312%20DEF.pdf. 10. Marine, I. Imtech Marine and CMAL win Electric & Hybrid Propulsion System and Supplier Awards. Available from: http://imtech.com/EN/Marine_EN-Newsroom/Marine_EN- Newsroom-Press-releases/(44780)-2014/Imtech-Marine-and-CMAL-win-Electric-and-Hybrid- Propulsion-Awards.html. 11. Marine, I. 38% Fuel savings achieved for CMAL hybrid ferry. Available from: http://imtech.com/EN/Marine_EN-Newsroom/Marine-Blogosphere/Fuel-savings-achieved- for-CMAL-hybrid-ferry.html. 12. EXPO, E.H.M.W. Liquid-cooled energy storage system development. Available from: http://www.electricandhybridmarineworldexpo.com/industry_news.php. 13. SLAVNICH, D. Electric pioneer. 2016; Available from: http://viewer.zmags.com/publication/061eea2c#/061eea2c/1. 14. Lorient, S.F., Supercapacitors versus batteries, in PASSENGER SHIP TECHNOLOGY. 2013, Riviera Maritime Media Ltd. 15. Lorient, S.F., Commuter craft prototype creates no emissions, in PASSENGER SHIP TECHNOLOGY. 2012, Riviera Maritime Media Ltd. 16. Deyrieux, F. Ferry powered by supercapacitors. 2013; Available from: http://www.s- lass.com/en/Documents/Presentationer%20Roadshow%20maj%202013/4_STX_2013%2005 %20Sweden%20ferry%20EL%20conference.pdf. 17. KNIGHT, S. Vision 2030. 2015; Available from: http://viewer.zmags.com/publication/3e159aac#/3e159aac/1. 18. KNIGHT, S. Legisation Outlook. 2015; Available from: http://viewer.zmags.com/publication/3e159aac#/3e159aac/1. 19. GL, D., DNV and Grieg Star take batteries On Board. 2013. 20. Contractors, H.M. Thialf. Available from: http://hmc.heerema.com/content/fleet/thialf/. 21. Xin, Q., Diesel engine system design. 2011: Elsevier. 22. Rakopoulos, C.D. and E.G. Giakoumis, Diesel engine transient operation: principles of operation and simulation analysis. 2009: Springer Science & Business Media. 23. Muenzel, V., et al. A Multi-Factor Battery Cycle Life Prediction Methodology for Optimal Battery Management. in Proceedings of the 2015 ACM Sixth International Conference on Future Energy Systems. 2015. ACM. 24. Fuchs, G., et al., Technology Overview on Electricity Storage: Overview on the potential and on the deployment perspectives of electricity storage technologies. Smart Energy for Europe Platform GmbH, ISEA-RWTH Aachen, 2012. 25. KNIGHT, S. Hybrid PSV Development. 2015; Available from: http://viewer.zmags.com/publication/3e159aac#/3e159aac/1. 26. Sen, P.C., Principles of electric machines and power electronics. 2007: John Wiley & Sons. 27. Linden, D., Linden's Handbook of Batteries. 4th ed, ed. T. B.REDDY. 2010. 28. Huggins, R., Advanced batteries: materials science aspects. 2008: Springer Science & Business Media. 29. Ferreira, B. and W. Van der Merwe, The Principles of Electronic and Electromechanic Power

101 | Page

Conversion: A Systems Approach. 2014: John Wiley & Sons. 30. Wang, J., et al., Combined state of charge estimator for electric vehicle battery pack. Control Engineering Practice, 2007. 15(12): p. 1569-1576. 31. Lu, R., et al., Analysis of the key factors affecting the energy efficiency of batteries in electric vehicle. World Electric Vehicle Journal, 2010. 4: p. 9-13. 32. University, B. BU-209: How does a Supercapacitor Work? [cited 2016 2016-02-29]; Available from: http://batteryuniversity.com/learn/article/whats_the_role_of_the_supercapacitor. 33. Ross, R., D. Stapersma, and J. Bosklopper, Fuel efficiency of diesel, electric and superconductive propulsion systems. Proc. INEC 10 th, Southampton, 2010: p. 367-380. 34. Kwasieckyj, B., Efficiency analysis and design methodology of hybrid propulsion systems. 2013, TU Delft, Delft University of Technology. 35. D.U. Sauer, G.B., A. Jossen, J. Garche, M. Rothert, M. Wollny, State of charge—What do we really speak about?, in Proceedings of the 21st International Telecommunications Energy Conference Presented at INTELEC. 1999: Kopenhagen. 36. Pop, V., et al., State-of-the-art of battery state-of-charge determination. Measurement Science and Technology, 2005. 16(12): p. R93. 37. Piller, S., M. Perrin, and A. Jossen, Methods for state-of-charge determination and their applications. Journal of power sources, 2001. 96(1): p. 113-120. 38. Charkhgard, M. and M. Farrokhi, State-of-charge estimation for lithium-ion batteries using neural networks and EKF. Industrial Electronics, IEEE Transactions on, 2010. 57(12): p. 4178- 4187. 39. Watrin, N., B. Blunier, and A. Miraoui. Review of adaptive systems for lithium batteries state- of-charge and state-of-health estimation. in Transportation Electrification Conference and Expo (ITEC), 2012 IEEE. 2012. IEEE. 40. Commission, I.E. White Paper Electrical Energy Storage. Available from: http://www.iec.ch/whitepaper/energystorage/?ref=extfooter. 41. Agency, I.E. Technology Roadmap Energy Storage. Available from: https://www.iea.org/publications/freepublications/publication/TechnologyRoadmapEnergyst orage.pdf. 42. ENERGY, E.C.D.-G.F. DG ENER Working Paper The future role and challenges of Energy Storage. Available from: http://ec.europa.eu/energy/sites/ener/files/energy_storage.pdf. 43. Kael Zach, H.A., Georg Lettner, Facilitating energy storage to allow high penetration of intermittent renewable energy. 2012. 44. Luo, X., et al., Overview of current development in electrical energy storage technologies and the application potential in power system operation. Applied Energy, 2015. 137: p. 511-536. 45. ULTRACAPACITORS, M.T. Product Comparison Matrix. Available from: http://www.maxwell.com/images/documents/Product_Comparison_Matrix_3000489_2.pdf. 46. Christen, T. and M.W. Carlen, Theory of Ragone plots. Journal of Power Sources, 2000. 91(2): p. 210-216. 47. Pell, W. and B. Conway, Quantitative modeling of factors determining Ragone plots for batteries and electrochemical capacitors. Journal of power sources, 1996. 63(2): p. 255-266. 48. Rahn, C.D. and C.-Y. Wan g , Battery systems engineering. 2013: John Wiley & Sons. 49. Hill, D.M., A. Agarwal, and B. Gully, A Review of Engineering and Safety Considerations for Hybrid-Power (Lithium-Ion) Systems in Offshore Applications. Oil and Gas Facilities, 2015. 4(03):

102 | Page

p. 68-77. 50. Hedlund, M., et al., Flywheel Energy Storage for Automotive Applications. Energies, 2015. 8(10): p. 10636-10663. 51. Hadjipaschalis, I., A. Poullikkas, and V. Efthimiou, Overview of current and future energy storage technologies for electric power applications. Renewable and sustainable energy reviews, 2009. 13(6): p. 1513-1522. 52. Zhou, H., Two-part modelling format for modelling marine diesel engines in joules, in Marine & Transport Technology. 2015, Delft University of Technology: Delft. 53. Patel, M.R., Shipboard propulsion, power electronics, and ocean energy. 2012: CRC Press. 54. MathWorks. Battery. Available from: http://nl.mathworks.com/help/physmod/sps/powersys/ref/battery.html?refresh=true. 55. Tremblay, O., L.-A. Dessaint, and A.-I. Dekkiche. A generic battery model for the dynamic simulation of hybrid electric vehicles. in Vehicle power and propulsion conference, 2007. VPPC 2007. IEEE. 2007. Ieee. 56. Rahn, C.D., Battery systems engineering. 2013, Wiley: Hoboken :. 57. TOSHIBA. SCiB Cells. Available from: http://www.scib.jp/en/product/cell.htm#cell02. 58. Energy, U.S.D.o., Grid Energy Storage. 2013. 59. Agency, I.R.E., BATTERY STORAGE FOR RENEWABLES: MARKET STATUS AND TECHNOLOGY OUTLOOK. 2015. 60. Deloitte, Energy storage: Tracking the technologies that will transform the power sector. 2015. 61. A R UP. Five minute guide to electricity storage technologies. Available from: http://publications.arup.com/Publications/F/Five_minute_guide_electricity_storage_technol ogies.aspx. 62. KNIGHT, S. Fuel Cell Future. 2015; Available from: http://viewer.zmags.com/publication/3e159aac#/3e159aac/1. 63. Carrasco, J.M., et al., Fast Response Energy Storage Systems, in Power Electronics for Renewable and Distributed Energy Systems. 2013, Springer. p. 367-427. 64. Sub-program, E.J.E.S.M.S., KINETIC ENERGY STORAGE BASED ON FLYWHEELS: BASIC CONCEPTS, STATE OF THE ART AND ANALYSIS OF APPLICATIONS. 2013. 65. energy, S. KINEXT 30/250 TECHNICAL SPECIFICATION SHEET. Available from: http://www.s4- energy.com/files/KINEXT_30-250.pdf. 66. Hansen, J.G. and D.U. O'Kain, An assessment of flywheel high power energy storage technology for hybrid vehicles. 2011. 67. Batteries, H.o. Lithium Ion - VL12V. Available from: http://www.houseofbatteries.com/lithium- vl12v-p-1145-l-en.html.

103 | Page

APPENDICES

A. EES energy capacity (for typical peak shaving function)

EES energy capacity in PS 3500 60 EES discharge/charge power EES cumulative capacity

2500 50

1500 40

500 30 Power(kW)

-500 20 Energy(kWh)

-1500 10

-2500 0 0 0.72 1.44 2.16 2.88 3.6 4.32 5.04 5.76 4 Time(second) 10 Figure 0-1 minimum EES energy capacity in PS (based on day1)

EES energy capacity in SB 3000 55 EES discharge/charge power EES cumulative capacity 2500 50

2000 45

1500 40

1000 35

500 30

0 25

Power(kW) -500 20 Energy(kWh)

-1000 15

-1500 10

-2000 5

-2500 0 0 0.72 1.44 2.16 2.88 3.6 4.32 5.04 5.76 4 Time(second) 10 Figure 0-2 minimum EES energy capacity in SB (based on day1)

104 | Page

EES energy capacity in AFT 3500 50 EES discharge/charge power EES cumulative capacity 3000 45

2500 40

2000 35

1500 30

1000 25

500 20 Power(kW) Energy(kWh) 0 15

-500 10

-1000 5

-1500 0 0 0.72 1.44 2.16 2.88 3.6 4.32 5.04 5.76 6.48 4 Time(second) 10 Figure 0-3 minimum EES energy capacity in AFT (based on day2)

EES energy capacity in PS 6000 90 EES discharge/charge power EES cumulative capacity 5000 80

4000 70

3000 60

2000 50

1000 40 Power(kW)

0 30 Energy(kWh)

-1000 20

-2000 10

-3000 0 0 0.72 1.44 2.16 2.88 3.6 4.32 5.04 5.76 6.48 4 Time(second) 10 Figure 0-4 minimum EES energy capacity in PS (based on day2)

EES energy capacity in SB 2000 30 EES discharge/charge power EES cumulative capacity

1500 25

1000 20

500 15 Power(kW)

0 10 Energy(kWh)

-500 5

-1000 0 0 0.72 1.44 2.16 2.88 3.6 4.32 5.04 5.76 6.48 4 Time(second) 10 Figure 0-5 minimum EES energy capacity in SB (based on day2)

105 | Page

B. Peak analysis

Another method which can directly analyze the power consumption measurements is used initially to derive design criteria for EES, however, it was not as good as the method used in 4.2.2 thereby it is only mentioned here in Appendix for readers who is interested.

Statistical data on the peak numbers, peak magnitude and peak duration can be obtained. A 15- minute time slot from AFT EES discharge/charge line based on day1 is selected to illustrate the quality of peak analysis. Analysis steps is explained below. Since discharge power is positive and charge power is negative as shown by discharge/charge line in Figure 0-6, therefore, the first step is to detach EES discharge line and EES charge line respectively from discharge/charge line and the result is plotted by green solid line and blue dotted line in figure below. Secondly, the inverted charge power line is created, which is illustrated by red solid line in the figure. The goal of first two steps is to secure fairly good accuracy of peak analysis.

EES discharge & charge 1000

800

600

400

200

-200

-400

-600

-800 EES discharge/charge power(kW) discharge power (+) ABS charge power (+) charge power (-) -1000 3.69 3.708 3.726 3.744 3.762 3.78 4 Time(second) 10

Figure 0-6 Detach result (based on AFT day1)

Definition of local peak in peak analysis is a data sample that is either larger than its two neighboring samples or equal to. Figure 0-7 is an example of find prominent peaks for EES discharge power of which the prominence is higher than 500 kW.

Find prominent peaks (EES discharge power) 1000 discharge power peak 800 prominence width

600

400 Power(kW)

200

0 3.69 3.708 3.726 3.744 3.762 3.78 4 Time(second) 10

Figure 0-7 Example of find prominent discharge power (based on AFT day1)

106 | Page

C. Overview technical data for EES

Table 0-1 (F-1) Overview technical data for EES

EES Energy Power density Specific Specific power density energy technology (W/l) (W/kg) (Wh/l) (Wh/kg) PHS 0.27-1.540 0.1-0.2 0.27-1.5 0.1-0.2 CAES 3-6 0.2-0.6 30-60 N/A VRFB 25-35 0.5-2 15-50 N/A HFB 55-65 1-25 75-85 N/A LA 50-80 10-700 30-50 75-300 NaS 150-300 120-160 150-240 90-230 NaNiCl 150-200 250-270 100-200 130-160 NiCd41 15-80 75-700 15-40 150-300 NiMH 80-430 500-3000 40-80 250-1300 Li-ion 200-640 1300-10000 60-240 150-2000 SMES 0.2-6 1000-4000 0.5-5 500-2000

FES 20-80 [40, 44] 600-5000 5-30 [40, 44] 400-1500 1-2042 1 to 1043

3 SC 1-6[45] (0.2-20)*10 [45] 1-15 (2-15) *103 10-20[40] (40-120)*103[40] Fuel cell 500-3000 500+ (0.8-100) *103 500+

40 Head 100m-head 550m, taking into account only the upper water basin. 41 Vented cells. Sealed cells are with higher energy density and specific energy. 42 See Appendix E. 43 See Appendix E.

107 | Page

Table 0-2 (F-2) Overview technical data for EES EES Response Depth of Energy Typical Operating technology time discharge efficiency lifetime temperature (°C) PHS 3 min44 80%-100% 75%-85% >15000 N/A CAES 3-1545 min 35%-50% 42%-54% 5000-20000 N/A VRFB < 0.5 ms46 100% 60%-75% >10,00047 10-40 HFB second 100% 65%-75% >200048 -30-50 LA 3-5 ms 10%-20%49 70%-75% (50-500)50 2551 NaS ms 100% 70%-85% 2500-4500 - NaNiCl < second 100% 80%-90% 1000 - NiCd < second - 60%-70% 1500-3000 -40-70 NiMH < second - 55%-65% 600-1200 -20-70 4 52 Li-ion

44 From maximum negative to maximum positive power output. 45 15 min from cold start. 46 Times for changing power depend very much on the actual operation condition of the redox-flow battery. E.g. if the battery is supplying high power the power can be changed to negative of the same power value without any delay. Only if the pumps are in stand-still it will take a few seconds until the power reaches full load in either charging or discharging direction. 47 At 100% DOD. 48 At 100% DOD. 49 Now advanced LA batteries can achieve DOD from 20% to 80%. 50 Up to 2000 cycles can be obtained with special designs, like deep cycle lead acid batteries: 100% DOD 150-200 cycles, 50% DOD 400-500 cycles; 30% DOD 1000 and more cycles. 51 25°C is the optimal operating temperature. Usable capacity of LA is highly temperature sensitive: the service life is halved with each 8°C rise. 52 Charge. 53 Discharge.

108 | Page

D. Estimation of practical values for SC

Maxwell supercapacitor products 6

3 Wh/l 2

0 0 5000 10000 15000 20000 W/l

Figure 0-8 Maxwell supercapacitor products (energy density and power density)

Maxwell supercapacitor products 8 7 6 5 4 Wh/kg 3 2 1 0 0 2000 4000 6000 8000 10000 12000 14000 16000 W/kg

Figure 0-9 Maxwell supercapacitor products (specific energy and specific power)

E. Estimation of practical values for FES

Regarding the mechanical design of a flywheel, things are simple since the energy is concentrated in the flywheel volume. A very good description of the mechanical design of the flywheel in given in [63, 64] and used here as following:

For a rotating disk, there are some useful and simple mechanical expressions that allow making interesting considerations on its size and speed. On the one hand, the kinetic energy stored in a spinning disk will be proportional to its mass times the square of the tip speed, while centrifugal stresses will be proportional

109 | Page

to the material density times the tip speed, hence the specific energy (per unit

mass “em” or volume “ev” ) can be expressed as:

= (1) 𝜉𝜉𝜉𝜉 𝑒𝑒𝑚𝑚 = 𝜌𝜌 (2)

𝑣𝑣 Where is the maximum stress𝑒𝑒 level𝜉𝜉𝜉𝜉 in the disk, its density and a form factor that only depends on the disk shape. For a cylinder its value is 0.6, for a 𝜎𝜎 𝜌𝜌 𝜉𝜉 ring 0.3 (even if it is a thick-wall cylinder) and 0.87 for an optimized geometry with the highest possible value for . Obviously, mechanical stresses in the flywheel during operation must be below the yield strength of the material. 𝜉𝜉 Equations (1) and (2) allow making considerations on how an optimum flywheel should be designed. First, it is important to distinguish whether volume or mass restrictions are more important. In general, for stationary applications volume is more a concern than the mass, while for moving applications mass optimization is mandatory. In any case, to achieve high energy densities, a high value of is required. Ring-shape flywheels should be avoided. Optimum-shape ones 𝜉𝜉 provide a high value for but are difficult to fabricate. Cylindrical flywheels are usually the preferable option. Regarding the material, there are also two 𝜉𝜉 choices: for high mass energy density, high strength and light materials should be used, while for high volume energy density, only the high strength of the material is a concern. Table 2 shows the mechanical properties of some selected materials and their ideal energy storage capability for a disc-shape flywheel. It can be inferred from Table 0-3 that the best choice for making an “energetic” and light flywheel is using carbon fiber, while using high strength steel will allow to make “energetic” and small machines but much heavier [63, 64].

Table 0-3 Mechanical properties of some selected materials (Source: [63]) Material (MPa) (kg/m3) (Wh/kg) (Wh/l)

Steel (AISI 4340) 1800𝜎𝜎 7800𝜌𝜌 39𝑒𝑒𝑚𝑚 304𝑒𝑒𝑣𝑣 Alloy (AlMnMg) 600 2700 38 102 Titanium (TiAl62r5) 1200 4500 45 203 Fiberglass (60%) 1600 2000 135 270 Carbon fiber (60%) 2400 1500 270 405

It should be noted that energy density and specific energy for FES in Table 0-3 is based on only the volume and weight of rotating disc. Actually, steel rotors have specific energy up to around 5 Wh/kg, while high speed composite rotors have achieved specific energy up to 100 Wh/kg. However, the energy density must be further reduced by a factor of 10 when consider to complete system weight (containment, vacuum system and electric interface) [64]. Thus the specific energy for FES is around 0.5 to 10 Wh/kg. And energy density can also be roughly estimated based on rotor specific energy and material density (because lack of data on rotor density), which is roughly from 39 Wh/l to 150 Wh/l. Further reduced by one order of magnitude,

110 | Page the energy density for FES system is then around 4 Wh/l to 15 Wh/l. To verify/modify the roughly estimated values above, a survey on FES products among main FES manufactures is shown in Table 0-4 below.

Table 0-4 Specifications for the FES systems (example1 to 4) Property Flybrid [50] S4 energy [65] Blueprint [64] Tribology[64]

Usable energy 111 Wh 30 kWh 0.75 kWh 1 kWh Power capability 60-110 kW 250 kW 120 kW 40 kW Max rotor speed 64,500 rpm - 28000 rpm 30000 rpm Rotor weight 5 kg 5000 kg - - System weight 25 kg 26500 kg 230 kg 126 kg System volume 18 l - 70 l 59 L Specific energy (rotor) 22.2 6 Wh/kg - - Specific energy (system)54 4.4 Wh/kg 1.1 Wh/kg 3.3 Wh/kg 7.9 Wh/kg Energy density 6 Wh/l 0.8 Wh/l 10.7 Wh/l 16.9 Wh/l

Table 0-5 Specifications for the FES systems (example5 to 8, Source: [64])

Property Powerthru CCM HyKinesys Magnet-Motor GabH Usable energy 667 Wh 4 kWh 300 Wh 2 kWh Power capability 190 kW 300 kW 100+ kW 150 kW Max rotor speed 52500 rpm 22000 rpm 28000 rpm 12000 Rotor weight - - - System weight - 375 kg 50 kg 400 kg System volume - 190 L 40 L 190 L Specific energy (rotor) - - - - Specific energy (system) 2.43 Wh/kg 10.6 Wh/kg 6 Wh/kg 5 Wh/kg Energy density 7.25 Wh/l 21 Wh/l 7.5 Wh/l 10.5 Wh/l

54 The flywheel system is defined here as including the energy storage rotor, the motor/generator, the bearings, and the containment. Power electronics weight, volume and cost are not considered.

111 | Page

Table 0-6 Specifications for the FES systems (example9 to 12, Source:[66]) Property Flywheel Ricardo University of Williams Energy Texas at Austin Usable energy 750 Wh 220 Wh 0.4-130 kWh 360 Wh Power capability 120 kW 60 kW 2-2000 kW 120 kW Max rotor speed 28000 rpm - - 36000 rpm Rotor weight - - - - System weight 150 kg - - 55 kg System volume 125 L 18 L - 38 L Specific energy (rotor) - - - - Specific energy (system) 5 Wh/kg - 5 Wh/kg 6.55 Wh/kg Energy density 6 Wh/L 12.2 Wh/l 10 Wh/l 9.47 Wh/L

From tables above, it can be seen that all the energy density data lies in the interval of from 1 Wh/l to 20 Wh/l (typically 10 Wh/l) and it is roughly consistent with the estimation of from 4 Wh/l to 15 Wh/l. And the system specific energy data from manufactures is around 1 Wh/kg to 10 Wh/kg (typically 5 Wh/kg), which also verifies the previous estimation of 0.5 Wh/kg to 10 Wh/kg.

F. Generator efficiency

Generator Efficiency(@nominal power factor) 0.98

0.97

0.96

0.95

0.94 efficiency1 vs. load1

Efficiency[%] NO.1-6 Gen. efficiency 0.93 efficiency2 vs. load2 NO.7-8 Gen. efficiency 0.92 efficiency3 vs. load3 NO.9-12 Gen. efficiency 0.91 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 Load[%] Figure 0-10 Generator efficiency

112 | Page

G. Results comparison (Excel & DE A4 static)

Excel Model Results DE fc Benchmark DE fc Hybrid Fuel savings Fuel savings deviation deviation (g) mode (g) mode (g) (g) (%) (%) DE1 (mak8m552) 3648025 5687434 DE3 (mak8m552) 4002853 5687434 DE5 (mak8m552) 2808041 DE7 (sulzer8zal40s) 2406552 PS day1 12865471 11374868 1490603 11.59% DE2 (mak8m552) 2916533 5738877 DE4 (mak8m552) 4013474 5738877 DE6 (mak8m552) 4188119 DE8 (sulzer8zal40s) 1761673 SB day1 12879799 11477754 1402045 10.89% DE9 (sulzer6zal40s) 3445055 5629031 DE10 (sulzer6zal40s) 0 DE11 (sulzer6zal40s) 743876 DE12 (sulzer6zal40s) 3363934 AFT day1 7552865 5629031 1923834 25.47% Overall1 33298135 28481653 4816482 14.46% 4840420.56 12.69% DE1 (mak8m552) 5165246 7155711 DE3 (mak8m552) 4969841 7155711 DE5 (mak8m552) 3241302 DE7 (sulzer8zal40s) 2480254 PS day2 15856643 14311422 1545221 9.74% DE2 (mak8m552) 5053612 6868532 DE4 (mak8m552) 4052989 6868532 DE6 (mak8m552) 3855116 DE8 (sulzer8zal40s) 2378826 SB day2 15340543 13737064 1603479 10.45% DE9 (sulzer6zal40s) 4231997 7255485 DE10 (sulzer6zal40s) 0 DE11 (sulzer6zal40s) 901605 DE12 (sulzer6zal40s) 4248418 AFT day2 9382020 7255485 2126535 22.67% Overall2 40579206 35303971 5275235 13.00% 4436754.65 9.86% Figure 0-11 Excel model results

113 | Page

Matlab/Simulink DE_A4 model Results DE fc Benchmark % (with DE fc Hybrid % (with Fuel savings Fuel savings deviation deviation (g) mode (g) Excel) mode (g) Excel) (g) (%) (%) DE1 (mak8m552) 3650631 0.07% 5570761 -2.05% DE3 (mak8m552) 4018819 0.40% 5570761 -2.05% DE5 (mak8m552) 2797366 -0.38% DE7 (sulzer8zal40s) 2458393 2.15% PS day1 12925209 0.46% 11141522 -2.05% 1783687 13.80% DE2 (mak8m552) 2898976 -0.60% 5622588 -2.03% DE4 (mak8m552) 4035285 0.54% 5622588 -2.03% DE6 (mak8m552) 4188360 0.01% DE8 (sulzer8zal40s) 1796119 1.96% SB day1 12918740 0.30% 11245176 -2.03% 1673564 12.95% DE9 (sulzer6zal40s) 3513530 1.99% 5585583 -0.77% DE10 (sulzer6zal40s) 0 DE11 (sulzer6zal40s) 749360 0.74% DE12 (sulzer6zal40s) 3436660 2.16% AFT day1 7699550 1.94% 5585583 -0.77% 2113967 27.46% Overall1 33543499 0.74% 27972281 -1.79% 5571218 16.61% 4595056.56 12.05% DE1 (mak8m552) 5126857 -0.74% 7028571 -1.78% DE3 (mak8m552) 4955916 -0.28% 7028571 -1.78% DE5 (mak8m552) 3214485 -0.83% DE7 (sulzer8zal40s) 2515707 1.43% PS day2 15812965 -0.28% 14057142 -1.78% 1755823 11.10% DE2 (mak8m552) 5026577 -0.53% 6735100 -1.94% DE4 (mak8m552) 4040555 -0.31% 6735100 -1.94% DE6 (mak8m552) 3858658 0.09% DE8 (sulzer8zal40s) 2420019 1.73% SB day2 15345809 0.03% 13470200 -1.94% 1875609 12.22% DE9 (sulzer6zal40s) 4313938 1.94% 7188220 -0.93% DE10 (sulzer6zal40s) 0 DE11 (sulzer6zal40s) 915385 1.53% DE12 (sulzer6zal40s) 4327645 1.86% AFT day2 9556968 1.86% 7188220 -0.93% 2368748 24.79% Overall2 40715742 0.34% 34715562 -1.67% 6000180 14.74% 4300218.65 9.55% Figure 0-12 DE A4 model results

114 | Page

H. High power Li-ion batteries

Table 0-7 High power Li-ion battery examples

Cell Max continuous Max continuous Type Life cycles Discharge charge

A123- 250C[27] 10C[27] LFP - AHR18700M1ULTRA 200055 Saft-VL12V 125C[67] 15C[67] NCA 500,00056 Toshiba-SCiB >10C[57] >10C[57] LTO >40,00057[57] (high power cell) Toshiba-SCiB 58 (high energy cell - >5C[57] LTO >15,000 [57] 20Ah)

I. List of abbreviation

AC Alternating Current) AFT After CAES Compressed Air Energy Storage CMAL Caledonian Maritime Assets DC Direct Current DCV Deepwater Construction Vessel DE Diesel Engine DG Diesel Generator DP Dynamic Positioning EES Electric Energy Storage EFD Energy Flow Diagram EGR Exhaust Gas Recirculation EMS Energy Management System ES Energy Source EU Energy Use FC Fuel Consumption FES Flywheel Energy Storage HFB Hybrid Flow Battery HMC Heerema Marine Contractors

55 >80% of initial capacity at 100% DOD. 56 3% DOD. 57 >80% of initial capacity at discharge/charge current: 10C/10C; SOC: 20%-80%. 58 >80% of initial capacity at discharge/charge current: 3C.

115 | Page

HTPEM High-Temperature PEM HV High Voltage HVAC Heating, Ventilation and Air Conditioning Systems JOULES Joint Operation for Ultra Low Emission Shipping KERS Kinetic Energy Recovery System LA Lead Acid Battery LCO Lithium Cobalt Oxide Battery LFP Lithium Iron Phosphate Battery LI-ION Lithium-Ion Batteries LMO Lithium Ion Manganese Oxide Battery LNG Liquefied Natural Gas LTO Lithium–Titanate Battery MCFC Molten Carbonate Fuel Cell MDO Marine Diesel Oil MOSFET Metal-Oxide Semiconductor Field-Effect Transistor NANICL Sodium Nickel Chloride Battery NAS Sodium Sulphur Battery NCA Lithium Nickel Cobalt Aluminum Oxide Battery NICD Nickel Cadmium Battery NIMH Nickel Metal Hydride Battery NMC Lithium Nickel Manganese Cobalt Oxide Battery OSV Offshore Supply Vessels PHS Pumped Hydro Storage PME Proton Exchange Membrane PMS Power Management System PS Portside PSV Platform Support Vessel RTG Rubber Tired Gantry SB Starboard SC Supercapacitor SFC Specific Fuel Consumption SMEC Superconductiong Magnetic Energy Storage SOC State Of Charge SOFC Solid Oxide Fuel Cell SPS SimPowerSystems SSCV Semi-Submersible Crane Vessel TLP Tension-Leg Platform UPS Uninterruptible Power Supplies VRFB Vanadium Redox Flow Battery

116 | Page