Overview and Initial Operational Experience of the ... - CiteSeerX

Proceedings of The Twelfth (2002) International Offshore and Polar Engineering Conference Kitakyushu, Japan, May 26 –31, 2002 Copyright © 2002 by The International Society of Offshore and Polar Engineers ISBN 1-880653-58-3 (Set); ISSN 1098-6189 (Set)

Overview and Initial Operational Experience of the LIMPET Wave Energy Plant Cuan B. Boake, Trevor J. T. Whittaker, Matt Folley Dept. of Civil Engineering, Queens University Belfast, Belfast, Northern Ireland

Hamish Ellen Wavegen Ltd Inverness, United Kingdom

Islay. It is situated close to the site of the now decommissioned 75kW prototype device that was installed by The Queens University of Belfast in 1991 (Whittaker et al, (1997). The new site is more exposed to the Atlantic Ocean (Mollinson, 1991) than the prototype unit and final commissioning commenced after the civil engineering construction was completed in September 2000. It has been designed as a research facility and as a demonstration unit supplying wave-generated electricity into the local grid. LIMPET is supported by the EU via the JOULE III programme. The initial objectives of the project were to:

ABSTRACT The 2nd full-scale shoreline Oscillating Water Column (OWC) wave energy device to be constructed in the UK, LIMPET, has been completed on the island of Islay, Scotland. The LIMPET device comprises a rectangular inclined OWC that ducts the generated airflow through two contra-rotating Well’s turbines, each coupled to a 250kW induction generator giving the device a 500kW maximum power output. This paper serves to provide an overview of the LIMPET device from conception to the current status and the initial operational experiences of running the plant.

• •

KEYWORDS: wave energy; LIMPET; Well’s Turbine; OWC; renewable energy.

• • •

Construct a shoreline OWC with an installed capacity of 500kW. Connect the plant to the local electricity grid and operate the plant as a prototype power station. Instrument the plant to monitor environmental loads, power train performance and the quality and quantity of delivered power. Experiment with different control settings to optimise the matching of the plant to different sea states. Compare full-scale performance with the predictions of mathematical and wave tank models.

The JOULE project partnership comprises: • • • • • Fig 1. The LIMPET shoreline OWC

Queens University, Belfast (QUB) (Project co-ordinator, development of collector form, turbine parameter specification, plant monitoring and research) Wavegen (Ireland) Ltd. (Plant owners and operators, design and manufacture of turbo-generation equipment) Charles Brand (Civil engineering contractors) Kirk, McClure & Morton (Civil engineering consultants) Instituto Superior Tecnico, Lisbon (Mathematical modeling of the system)

The project partners form a good blend of industrial companies, broadly responsible for the construction and operation of the plant, and academic institutions responsible for the conceptual development and research activities.

INTRODUCTION The LIMPET (Land Installed Marine Powered Energy Transformer) shoreline OWC (Sarmento and Falcao, 1985) is a 500kW wave energy collector located on the South Western coast of the Hebridean island of

586

research on alternative power take-off systems operating simultaneously and independently, and on the full-scale hydrodynamic behaviour of OWC’s in varying configurations.

Design Overview OWC Form The LIMPET OWC is inclined at an angle of 40o to the horizontal (Fig. 2). This has two distinct advantages in relation to a vertical water column such as used on the 75kW prototype: •

•

The structure was designed to withstand either a frontal wave pressure of 6 bar across the full 21m width of the device, or a peak internal pressure of 1 bar. To avoid the uncertainties associated with rock quality at the site, a rear wall was added to the structure so that the internal pressure forces were totally contained within the collector structure. In this way the requirements for anchoring the structure to the site rock was minimised.

The inclined column and large radius front-lip reduces entrance turbulence and internal sloshing (Muller and Whittaker, 1995). This is particularly true at the shoreline where shallow water effects increase the surge motions relative to heave. The inclination of the water column increases the water plane area of the water column for a given chamber cross section. The entrained mass of water determines the water column resonance and which can be coupled to the predominant period of the incoming waves.

Independent tank tests by both QUB (McStay, 1995) and Wavegen confirmed the improved hydrodynamic performance of inclined water columns in comparison to their vertical counterparts. Fig 3. Collector Cross Section Air pressure from all three water columns is equalised by 3m x 2.4m openings in each diaphragm wall at bench level. These openings can be blanked off to achieve individual column isolation. Air exits the collector through one of two 2.6m diameter circular openings in the back wall and passes into the turbo-generation duct. The central opening will be used at the time of initial commissioning. The second opening will be blanked off but may be used at a later date to test alternative equipment. One metre square openings have been left in the roof of two collector chambers to allow for retrofitting of pressure relief valves as part of the control strategy. Both these orifices are blanked off at the time of initial commissioning.

Fig 2. Axial Section through collector The wave breaker (Fig. 2) on the collector reduces the amount of water overtopping the collector and any water falling behind the collector is highly aerated rather than of high density.

Turbo-Generation System And Controls The operational design parameters for the Wells turbines (Gato, Warfield and Thakker, 1993) to be fitted to the collector were specified by QUB and the responsibility for the design and construction of the turbo-generation equipment conforming to these parameters lay with Wavegen.

A horizontal bench at the top of the column inclined rear wall is a safety feature to reduce the likelihood of entrained water flowing into the turbine duct. Should the water level attain bench height, it would still have a significant height to rise before reaching the turbine axis.

During the operation of the 75kW prototype device, the team at QUB developed a significant database on the energy incident of the prototype test site (McIwaine, 1992). Through a detailed analysis of this information, a set of 53 sea states was developed as representative of the wave climate. From this data an annual average incident wave energy of 17.9kW/m was estimated (Whittaker et. al, 1997). The reference location for the source data is some 300m from the actual LIMPET site and is relatively sheltered and as such it was expected that the actual power incident on LIMPET would exceed the estimated value.

The external front lip section of the front wall steepens to an angle of 60° to the horizontal. This effectively throttles the water inlet/outlet area and is an important feature in minimising inlet broaching. A phase lag between the internal water level and the external water level during outflow ensures that there is continual flow of water out of the column even when the external water level is below the entry lip. The cross-section of the collector (Fig. 3) is divided into three separate columns for several reasons: As the width of the column increases there is an increasing risk of transverse wave excitation within the water column. This reduces the energy capture performance of the column.

It is difficult to judge the optimal size of a LIMPET type device but it was considered that the next stage of development after the 75kW prototype should represent a significant size increase and offer the basis for the modular development. An installed capacity of 500kW and a utilisation of 40% to give an annual average output of 200kW (Whittaker et. al, 1997) appeared to be reasonable targets.

The depth of roof required to span the 21m width of the column without additional support is so large as to be economically inefficient. The isolation of individual columns is possible; this permits future

587

Testing at both QUB and at Wavegen in Inverness indicated that correctly tuned OWC/Wells turbine/induction generator systems should offer an overall conversion efficiency to electricity of 50% (Curran, Whittaker, Beattie and Raghunathan, 1998) of the power incident on the collector width and on that basis an overall collector width of 21m was selected as suitable to meet the power output objectives.

Perc ent

40

Table 1. Turbine parameters Turbine Diameter

35

% Energy Contribution

30

% Occurrence

25 20 15 10

2.6m

5

Nominal Operating Speed 1050rpm

0

Number of Turbines

2

Arrangement

In Line Contra-rotating

Blade Form

NACA12

Number of blades

7

Blade Chord

320mm

Hub to Tip Ratio

0.62



Output Power (k W)

Fig 6. LIMPET Model Trials – Power Output made via a short duct section. There are two in-line valves between the collector and the first of the two 250kW turbo-generators. Two valves are used because of the safety critical nature of their function. The valves are of differing design and use different actuation methods to minimise the risk of common mode failure.

A series of 1:40 scale tests of the proposed device, the immediate local coastline and the perceived bathymetric profile were conducted in tank trials at the Wavegen test facility in Inverness (Boake, 1998). A wave to wire modelling technique (Curran, Whitttaker and Stewart, 1998) was employed to predict the annual average electrical power for various turbine configurations and resulted in the specified turbine parameters (Table 1). The performance predictions of the OWC and turbines are shown in Figures 4-6. 70 % Energy Contribution % Occurrence

60

Percent

50 40 30

Fig 7. Ducts and Mechanical Plant

20

The assembly consists of an entry duct section (1, Fig.7), an isolating butterfly valve (2, Fig.7), a second duct section (3, Fig.7) and a further isolating radial vane valve (4, Fig.7). The Turbine/ Generator assemblies are mounted opposite each other in the airflow to provide a contra-rotating turbine system. Each generator (5,7, Fig.7) has a through shaft which on one end has the turbine arrangement and on the other has a large flywheel. The two turbines are separated by a turbine runner (6, Fig.7) and the airflow is drawn in and expelled through a bell-mouth (8, Fig.7). Each turbo-generation module consists of an induction generator mounted on a bespoke steel frame. The generators are specifically designed such that their bearings will take the combined axial and radial loads imposed by a directly mounted turbine. A Wells turbine is mounted at one end of the generator shaft and a large flywheel at the other. The two modules are fixed back to back so that the combined effect is that of a contra-rotating bi-plane unit.

10 0

Average Pneumatic Power (kW)

Fig 4. LIMPET Model Trials – OWC Pneumatic Power

100.00 90.00 Turbine Efficiency

80.00 70.00 60.00 50.00 40.00 30.00 20.00

Each of the generators is inverter driven so that advantage may be taken of the large system inertia (Falcao, Whittaker and Lewis, 1994). The inverter drives require either a torque or a speed reference signal from the master control system and it is an important part of the total LIMPET programme to develop and test alternative control strategies.

10.00 0.00 0

500

1000

1500

2000

2500

3000

Average Pneumatic Power (kW)

Fig 5. LIMPET Model Trials – Turbine Efficiency

Work at QUB (Alcorn, Beattie and Cully, 2000) has shown that pressure relief valves can improve the overall efficiency of LIMPET relative to the baseline control system using in line valves. A relief valve is to be fitted to LIMPET to give a field comparison.

The collector is fitted with two outlets. In the initial stages of operation one outlet will remain closed whilst the second is connected to the turbo-generation equipment. The turbo-generation duct connects to a circular outlet in the back wall of the collector, the connection being

588

increased until the maximum is reached (the maximum available being determined either from grid or generator limitations). If the turbine speed falls below a set minimum the demand torque is zero. This prevents the turbine falling to a low speed from which it cannot absorb sufficient power to recover.

Electrical Equipment The electrical system is made of standard components suitable for installation in the harsh marine environment. The system is comprised of a simple switchboard feeding each generator system and connecting the plant to the grid via a 400/11000V transformer. Each generator is controlled by the supervisory plant controller that in turn controls an inverter drive connected to the machines. Each generator is driven by an inverter drive that is controlled in torque mode. The torque demand is supplied to the inverter controller by an algorithm developed by Wavegen which is incorporated into the supervisory microprocessor control unit. The inverters are fully rated for the 250kW and control the stator of each induction machine. Aside from the generators all the electrical equipment is housed in a building 20m remote from the turbine arrangement.

A separate but identical algorithm controls each generator. The inverter drives provide a valuable level of flexibility to the system allowing the turbines to be operated at variable speed ranges depending on the incoming power levels. Currently this process is undertaken manually adjusting the range of operating speeds as the incoming power level changes. This process is typically of a daily or weekly nature. An automatic algorithm which will adjust the operating speed range according to the average incoming power on a minute-by-minute or hour-by-hour nature is under development.

Control System

Valve Position Control

The supervisory control unit has three generic functions:

The function of the two in line valves (vane and butterfly) is to reduce airflow to the turbines in storm conditions and to close in an emergency. The position of the control valve (which could be either valve, currently vane) is also based on speed such that if the turbine is running beneath a first set speed the valve is fully open and between this and a second set speed the valve closes linearly to zero.

•

The controller determines whether it is safe and desirable to operate the plant. It controls the starting of the machine. It controls the generation and operation of the plant initiating an appropriate shutdown procedure in the event of problem.

• •

OWC Construction

Before starting the plant the system performs a number of checks including: • • •

With the exception of the lower section of the roof structure, the collector was built primarily of in-situ cast reinforced concrete. The lower section of the roof was formed from beams pre-cast at site and which provided a firm foundation for an in-situ capping without the need for complex scaffolding.

E-stop circuit closed. No warnings from any monitoring equipment Adequate energy entering the water column

While running the plant produces power under the control of the Wavegen algorithm until either a fault condition is detected or wave activity falls to a level that is insufficient to sustain generation. The control algorithm provides an independent torque reference for each of the two generators and also a position demand signal for the modulating valve. The demand torque is currently chosen by an algorithm based on the speed of the turbine, so as the turbine speeds up the torque demand is

Fig 9. Construction Concept of LIMPET Demand Torque (%)

Vane Valve Position (Degrees)

The major problem faced when building the collector was the protection of the construction site from wave activity during the construction phase. This was achieved by excavating a hole behind the existing cliff edge and building in the lee of the protective rock bund (Fig. 9). During adverse weather conditions, waves overtopped the bund and it was necessary to cease working at the base of the structure. However, the degree of protection was sufficient to limit lost time to 25% during the summer months when construction activity had been planned. In general terms the construction team was able to use 10-day weather forecasts to predict weather down time and could plan accordingly.

120 100 80 60 40 20 0 700

800

900

1000

1100

On completion of the excavation, construction commenced with the casting of the rear wall. This was followed by the erection of the sidewalls and finally the roof of the structure to leave the structure complete but isolated from the sea by the wave wall (Fig. 9).

1200

Generator Speed (rpm)

Fig 8. Control Algorithm

589

The final stage of the construction was then to demolish the wave wall by using explosives. The rock was overcharged in relation to a normal quarrying operation in order to ensure that the shattered rock was in small enough pieces to be removed by a long reach excavator.

control of the loggers via the Ethernet local area network (LAN). Included on this network is a video server that controls a maximum of four video cameras and streams live video images onto the LAN. The video cameras are used to monitor column movement, sea-state conditions and potentially for streaming to the Internet.

Prior to blasting the wave wall, butterfly valves had been fitted to the two outlets on the landward side of the collector to facilitate the fitting of power take off systems. The turbo-generation equipment, which had previously been assembled and tested by Wavegen was transferred to the LIMPET site and connected to the collector. After installation, the turbo-generation equipment was enclosed in a simple building to give protection against the worst vestiges of the weather.

It is important to note that the plant controller system operates entirely independently of the data acquisition system. However, an interface between the two disparate systems was devised to permit synchronised data flow between the two systems. Four ISDN2 connections in the control room provide remote dial-up access to the system for authorised download of data and video images, real-time monitoring and remote control of the plant.

Construction Summary Overall the construction took longer than originally planned. The original proposal scheduled construction completion within a single season; however in the first year the contractors ran into major difficulties not directly associated with the project. Consequently relatively little progress was achieved in the first year. During the second year, however, Charles Brand demonstrated that with correct preparation, it was possible to build effectively in the prevailing conditions and that a LIMPET style device could be constructed in a single season. Significant lessons have been learned during the building of the structure of LIMPET and these are already being applied to the design of successor devices.

C

W

10BaseT Ethernet Network

Waveloading Beams Interface to Controller

1

The incoming wave climate at a site is arguably the most difficult of all the plant parameters to monitor. Commercially available systems are expensive and tend to provide statistical data only via wave buoys moored a safe distance offshore. There has always been a requirement to acquire time-series data of the incoming waves close to the device itself to investigate the device response and performance fully. The OWC responds to a wave-by-wave excitation and a statistical description of the wave climate regime is sufficient mainly to describe the design space the device should fall within. Additionally, real-time time-series wave amplitude data can be used for real-time device control to optimize productivity.

Remote Clients Remote Link ISDN8

• High speed scanning • Triggered Burst Mode

HOST SERVER COMPUTER •Win NT Server •WWW Server •Overall Control of Client Applications and outside world connectivity • Data processing/analysis/archiving • Data backups

RS - 485

Remote Control • PC - Anywhere

datalogger

Remote Access

•Win NT Clients • Intranet

• Stand alone datalogger Video Server

Coaxial Composite Signal Cable

• TCP/IP Protocol to host • Interface to Controller

General DAQ/Monitoring

W. C.

2

Seabed Pressure Transducers

datalogger

• Embedded signal conditioning

Sea-bed Pressure Transducers Diaphragm Wall Pressure Transducers Ultrasonic Units Chamber Pressure Chamber Temperature Chamber Video Surveillance Duct pressures Atmospheric Pressure

Fig 11. LIMPET Instrumentation Coverage

LIMPET DAQ Monitoring & Control System

• Stand alone datalogger

• 5 Hz scanning

1. 2. 3. 4. 5. 6. 7. 8.

Fig. 11 depicts the regime of plant parameters that are monitored and these are discussed in the following sections:

Waveloading Modules

Plant Controller Unit

6

7

System Overview

• TCP/IP Protocol to host

5

3

Instrumentation and Data Acquisition

• Embedded signal conditioning

4

8

Video Camera 1

Video Camera 2

Video Camera 3

Data Retrieval •PC - Anywhere Autotransfer

A stipulation for real-time data acquisition presents the biggest problem for any remote instrument in a harsh environment; namely the power source for the instrument and less importantly, data storage. Indeed, it is these factors that constrain most remote systems to statistical observations in order to conserve battery power.

Marketing •WWW •Selective Data Visible

Fig 10. LIMPET DAQ Monitoring & Control System

QUB has extensive experience in monitoring the waves created by fast ferry vessels in coastal waters. A portable battery-powered system was developed that measures the static head of water above the seabed and which is then calibrated to provide the time series of wave amplitudes passing overhead (maximum battery power 4 hours, maximum data storage 3 hours @ 2Hz). A similar system was developed for LIMPET with the exception that the undersea rig housing the pressure transducer would be connected to the shore via an umbilical pipe fixed to the seabed by rock-anchors. Stainless steel modules were developed that housed the pressure transducer, an expandable outer rubber locking diaphragm and the cable connections. The cables consisted of a power cable, data cable, airline to the diaphragm and a steel cable for

The LIMPET device was designed to accept an extensive range of instrumentation. An Ethernet network comprises the backbone of the Data Acquisition and Management System. Two dataloggers form the core of the data acquisition system and are housed in a cabinet at the collector rear wall and enclosed by the turbo-machinery building. One datalogger is configured for high-speed burst-mode scanning for wave-loading data acquisition and the second logger operates at 5Hz and acquires plant operational data from various sources. All instrumentation cables terminate at the cabinet housing the loggers. A server computer situated in the control room 50m away from the device handles the remote

590

retrieving the module. The umbilical pipe would thus provide a reusable passage for the modules to be transported to the seabed rig by compressed air. Once in position at the rig, the modules would be locked into position by pressurizing the outer bladders. Pressure in the locking diaphragms would be maintained by a bank of pressurized cylinders. The cylinders’ pressure would be electronically monitored and maintained by a small air compressor. Laboratory trials with the system proved very successful. However, the system relied inherently on the use of divers for deployment on site and an accompanying weather window. Local divers on Islay were contracted to install the rigs and pipes to minimize the logistical overhead in bringing divers from Belfast. Disappointingly, the divers failed to meet their obligations despite several suitable weather windows, and it became apparent that an alternative solution would have to be devised before the onset of winter storm activity.

winter months where the relative humidity of the air is higher. Pressure transducer readings are unaffected by this vapour while ultrasonic signals are again vulnerable. Access to the ultrasonic units for maintenance or installation requires favourable weather conditions while pressure transducer maintenance is independent of weather conditions. Water Column Displacement about local datum 3.5 Ultrasonic Pressure Transducer 3

height (m)

2.5

New pyramid-shaped rigs were constructed from rolled steel joists (RSJ) of sufficient weight to eliminate dragging and the umbilical pipes were replaced by armoured cable weighted with galvanized chain. The cables and rigs were deployed from a local fishing boat at distances of 44m and 66m from the front lip of the device in an operation that took only 3 hours. Critically, the researchers from QUB were able to identify the favourable weather window and to mobilize the resources to site within 4 hours. This highlights the desirability of maintaining control of such tasks in-house; the reliance on local sub-contractors in projects of difficult (or remote) location introduces an additional complexity to the likelihood of success or failure in the task.

2

1.5

1

0.5

0 time (s)

Fig 12. Comparison of Water Column Displacement To date, these instruments have performed reliably and Figure 12 shows the close correlation between the systems. There is a discernible phase lag between the ultrasonic signal and the pressure transducer signal. The ultrasonic units employ some real-time statistical features that incur some processing overhead that results in the slight time delay. This time constant was established from laboratory tests to be approximately 2 seconds; corrected ultrasonic readings and the pressure transducer readings appear to track the column movement accurately.

The critical zone in this system is the air-water interface where the cable is routed from the seabed to land. No obvious route to land on the shoreline existed. Ideally the cable would have been routed through the device chamber entrance and up the rear wall, thereby avoiding the extremely aggressive turbulent area in the gully during storms. Deterioration of the weather eliminated this possibility and the cable/chain was routed up the Northern side of the gully wall and up the front wall of the device.

From an operational perspective, the ultrasonic units require favourable weather windows for installation or servicing as access to the front wall of the device is required. The pressure transducers can be retrieved at any time for replacement or servicing. As with all the instrumentation, however, adequate provision for access ducts must be detailed prior to and during construction. In this instance, ducts for the pressure transducers were specified for all the longitudinal diaphragm walls of the chambers for redundancy. This decision was justified when some of the ducts were found to be impassable by the pressure transducer modules; due most likely to poor routing of the ducts or distortion of the ducts during concrete pouring.

Approximately two weeks of data was acquired from the two rigs before the cable of the 44m rig was destroyed during storm conditions. A further two months of data was acquired from the 66m rig before it too was destroyed. In both instances, the damage occurred at the front wall where the cable and chain were subjected to the severest fatigue loads. However, sufficient data was acquired in various sea-states to permit analysis of the device response to specific wave excitation. Importantly, valuable operational experience was gained from the deployment of this system and it is envisaged that new cables will be deployed in the summer 2002 and routed up the rear wall of the device.

Chamber and Duct Pressures Pressure transducer units were installed in the rear walls of the chambers. The turbo-generation duct was comprehensively fitted with pressure transducers:

Water Column Displacement In addition to ultrasonic units positioned in the roof of the collector, wave column displacement is measured by pressure transducers that record the static head of water in the chambers. These transducers are located at the bottom of the diaphragm walls via ducts that run the length of the diaphragm walls and exit at the rear wall of the device. The pressure transducers are of the same design as described in the previous section and thus may be retrieved should failure occur. The pressure transducers are advantageous in that the recorded head inherently integrates across the surface of the water column, thereby minimising the effects of sloshing and uneven surface plane area. Ultrasonic signals are vulnerable to erratic behaviour because of the latter factors. Additionally, video footage of the chamber has revealed the occurrence of water vapour that appears in the chambers once the pressure drops below a certain threshold and especially during colder

• • • • •

between the butterfly valve and vane valve (pos. 3, Fig. 4) between the vane valve and first rotor (pos. 5, Fig. 4) between the turbine rotors (pos. 6, Fig. 4) after the 2nd rotor and before the bellmouth (pos. 7, Fig. 4) in the acoustic baffle room

The progressive train of pressure transducers has performed reliably to date. Chamber Temperatures Incorporated into the chamber pressure units of the southern and central chambers are temperature sensors. While these appear to have

591

performed reliably, preliminary analysis of thermodynamic models has predicted higher temperature swings than those recorded by the thermocouples. This is possibly due to the moisture content of the air within the chambers.

Interface to Controller The plant controller system (operating completely independently) monitors numerous plant operating parameters not included with the data set as acquired by the dataloggers. Thus a RS-485 interface and communication protocol between the plant controller system and the main datalogger was devised to permit synchronised data acquisition of all the relevant plant parameters. An instantaneous snapshot of the plant operation would thus be available at 5Hz. The following parameters were supplied by the controller to the datalogger:

Waveloading Two waveloading beams were fitted to LIMPET; an external beam on the lower front lip of the device and an internal beam on the rear wall of the device (at mean water level). This operation required very calm sea conditions for access down the front wall and into the chamber. Two of the authors, Boake and Ellen, achieved a Level 1 qualification in Industrial Rope Access to comply with Health and Safety regulations for such access.

• • • • • • • •

The beams were successfully installed and have been operating reliably to date. The signals from the pressure transducer arrays are routed to a dedicated datalogger. In the initial phase of monitoring, the datalogger was configured to operate in statistical mode whereby the following parameters were reported every hour:

This interface operated reliably apart from instances of high turbine acceleration when data “dropouts” would occur. No data string would be received by the datalogger from the controller before the 5Hz timeout occurred. It is possible that signal noise was responsible for disrupting secure transmission between the systems; this has since been addressed with a revised interface protocol.

average pressure standard deviation of pressure maximum pressure time and date of maximum minimum pressure time and date of minimum number of samples

Speed (rpm)

• • • • • • •

During severe storms in January 2002, pressures of 4.5 bar were recorded on the front wall of the device. To validate these maximums, the datalogger has since been configured for triggered burst mode operation at 1kHz scan speed. Time series traces of the pressure readings at the front wall during storm conditions will reveal whether these high recordings are feasible.

Generator 1

950

Generator 2

900 850

Fig. 13 Generator RPM – supplied by Plant Controller

Output Power (W)

Video Surveillance Two CCTV analogue video cameras were installed on LIMPET:

•

1000

Time

Pressure readings from the internal beam have not been significant.

•

Grid voltages - 3 Phases RPM Generators 1 & 2 Chamber Pressure (independent) Butterfly Valve Position (demand & actual) Vane Valve Position (demand & actual) Power Factor Generators 1 & 2 (demand & actual) Output Power Generators 1 & 2 Torque Demand Generators 1 & 2

100000 80000 60000 40000 20000 0 -20000

Generator 1 Generator 2

Time

In the central chamber rear wall and angled downwards to observe the water surface motion. Illuminating spotlights were installed at the outer sides of the rear wall to illuminate the interior. On a 3m mast 30m to the south of the device and providing a view of the sea and gully.

Fig. 14 Power Output - – supplied by Plant Controller An essential parameter supplied by the controller is the plant power output (Fig. 14). It can be seen that the output of Generator 2 is approximately half that of Generator 1 although the corresponding generator speeds (Fig. 13) are within a few RPM of each other. Since the torque demand to each generator is speed dependant, this is an unlikely scenario and thus it has not been possible to analyse the power train from pneumatic power to electrical power with any degree of confidence. It is proposed to perform independent power measurements to investigate this discrepancy.

Both videos were connected to a video server that converts the composite video signals into the TCP/IP protocol for Ethernet transmission. The video server is allocated an IP (Internet Protocol) address; remote clients logging in to the host server computer are then able to view the video images in a local Internet browser courtesy of the intranet connection. The quality of transmitted images to a remote ISDN2 connection was of an acceptable standard with high refresh rates.

Remote Communications

Video streaming from the column video gave a valuable insight into the characteristics of the column motion and which was compared to a model scale investigation into the flow characteristics of LIMPET (Folley and Whittaker, 2002).

This is an essential aspect of the plant operation in lieu of the remote device location. After a period of instability, the host computer system and associated telecommunication equipment has performed reliably. The following system specification proved to be the most reliable: •

592

Host Computer running WIN 2000 Server.

• • •

by nearly 50%. This is a consequence of the behaviour of waves in shallow water. As water depth reduces in relation to wave height waves increasingly “feel the bottom”. Friction between the moving water particles and the seabed causes a loss of energy in the wave and also changes the wave profile from sinusoidal to cnoidal. The reduced water depth over a relatively long distance both reduces the incident wave power and creates significant distortion to the sinusoidal waveform for which the collector was designed.

Remote Windows Clients login via dial-up connection into an ISDN2 Terminal Adaptor. Automatic reboot at midnight every night eliminates lockouts of more than 24 hours. Remote control of server using a Symantec PC-Anywhere TCP/IP session over a dial-up connection. This allows upgrading of software or data transfer from the host computer.

Operational Experience

Gully Shape

During a severe storm shortly after commissioning the plant in December 2000, the plant output reached the grid limit of 150kW and valve modulation occurred as planned. After this encouraging start the output of the plant appeared to fall significantly and it soon became clear that the disappointing performance was not simply attributable to low energy sea conditions. This was a possibility in that islanders have reported that in general the period 1999-2001 has seen unusually mild sea conditions. The change in plant performance was symptomatic of a blockage of the collector mouth and a diver survey was commissioned to check on this. In respect of local sea conditions the earliest that the survey could be made was mid March but the survey did show that there was an accumulation of broken rock at the collector entrance which was causing a blockage estimated at more than 75% of the entry area. An examination of the rock showed clearly that it was residue from blasting the wave wall and was unrecovered rock that had washed back into the gully during the winter. A longer reach excavator (22m) was taken to site and the residue cleared. A further diver survey confirmed that the entry lip, the gully and the area for some way out from the gully is now substantially clear of debris.

The LIMPET structure sits at the end of a man made gully with straight sides and a length of approximately 17m. The productivity testing in the wave tank was performed on a model gully with a flare angle of 12.5o on each side. Research at QUB (Stewart, 1993) has shown that the capture performance of the OWC varies with the factor

w w + 2 * l * sin(θ )

where w is the collector width, l the gully length and θ the flare angle. Thus the capture of the parallel gully is less than 75% of the tested flare performance. As a consequence of constructional difficulties it became necessary, to allow completion of the construction within the summer season of 2000 to proceed without the flare. This was however at the expense of performance. The combination of survey error and gully form serves to reduce the predicted output of the plant to less than 40% of the original expectation. We are still at the early stages of plant monitoring and it is not appropriate to draw full conclusions at this stage. It is likely however that on the basis of observations that even during the winter months the output will fall short of expectation and that the effects of lower water levels than expected and absence of flare will reconcile full scale performance to model predictions.

With the entry clear the plant performance improved significantly but still does not meet the original expectation. On the basis of the performance projected at project initiation we would have expected average generation to be approaching the 150kW grid limit during September and October. In practice the output was less than one third of this. The reasons are two fold with both relating to site topography.

If the initial performance data is confirmed then it will be a considerable disappointment. It should however be stressed that the reduced output does not suggest that there is any fundamental problem with the principle of the OWC or the ability to develop the technology to the commercial stage but rather that there was not enough emphasis given to checking the detailed basis of design for LIMPET before the start of construction. We should also take heart that when comparing like with like the performance of the device at the full scale will replicate that indicated by model tests. This will give a major confidence boost to the designers of the next generation of shoreline devices.

Sea Floor Topography The predicted output of the device was based upon model tests using 53 wave spectra derived by QUB as representative for the site of the 75kW prototype device. The water depths and seabed profile modelled in the tank testing were based upon an early survey. Applying the test data to the available data on turbine efficiency an annual average output approaching 150kW was estimated. In moving to the more exposed location of LIMPET it was estimated that the incident wave power might be as much as 30% higher than at the prototype site and as such it seemed reasonable to predict a potential annual average output of over 200kW. The grid capacity at Portnahaven limits the plant output to 150kW and this has the effect of dropping the predicted annual average to 111kW.

Operational Reliability To date the plant has proved extremely reliable and there have been no major mechanical or electrical problems. Since May 2000, save for periods allocated to research, the plant has been running under full automatic control with remote control from the Wavegen offices in Inverness. A total of 1800 hours of generating operation have been achieved out of a possible 4000 hours (grid OK and plant OK). During this time there have been numerous shut downs which have been caused either by a reported fault or by a decline in wave activity. Of the reported faults approximately half have been a consequence of false signals from the instruments and half due to local grid faults. In the longer term the outage due to local grid faults gives cause for concern and the lack of stiff grids at suitable sites for wave energy plant is one of the major barriers to development. We are nonetheless greatly encouraged by the relative lack of operational difficulty to date.

When the site was resurveyed during construction it was found that there were significant differences between the early survey and what was found on the ground. The two principal differences being: 1. 2.

(1)

The water level at the cliff edge was more than 1m less than expected. The seabed slope of 1:25 described on the original survey as starting at the cliff edge did not in fact start until some 60m away from the cliff.

Whilst these differences seem small the effect on energy capture as determined by model tests is dramatic with the captured power falling

593

Curran, R., Whitttaker, T.J.T., Stewart, T.P., Aerodynamic Conversion of Ocean Power From Wave to Wire, Jour. Energy Conversion & Management, Elsevier Sci., Vol. 39, No. 16-18, pp 1919-1929, 1998. Gato L.M.C., Warfield, V., Thakker, A., (1993). Performance of a high-solidity Wells turbine for an OWC wave power plant, European Wave Energy Symposium. McIwaine, S.J., (1992). An investigation of the performance of the Islay shoreline wave energy conversion unit, Ph.D. thesis, The Queen’s University of Belfast. Falcao, A.F., Whittaker, T.J.T., Lewis A.W., (1994), European Pilot Plant Study, Joule 2, Preliminary Action: European Pilot Plant Study, Contract No JOUR-CT91-0133 Report, 11Opgs. Folley, M., Whittaker, T.J.T., Identification of non-linear flow characteristics of the LIMPET shoreline OWC, Proc. 12th International Offshore and Polar Engineering Conf., Kyushu, Japan, 2002. School of Civil Engineering, Queen’s University Belfast, Northern Ireland. Mollinson, D., (1991). The UK wave power resource. I. Mech. E., Wave Energy Conference, London. McStay, P., (1995). An Experimental Study of Cylindrical Oscillating Water Column Wave Power Devices, MSc. Thesis, The Queen’s University of Belfast. Muller, G., Whittaker, T.J.T., (1995). Visualisation of flow conditions inside a shoreline wave power-station, Ocean Engng, Vol 22, No. 6, pp. 629-641, Elsevier Science. Muller, G.U., Whittaker, T.J.T., (1995), Field Measurements of Breaking Wave Loads on a Shoreline Wave Power Station, ICE Proc. Water Maritime and Energy, Sept. 1995, pp 187-197, ISBN: 09650946. Sarmento, A.J.N.A., Falcao, A.F. de O., (1985). Wave Generation by an Oscillating Surface-Pressure and its application in Wave Energy Extraction. J. of Fluid Mech., Vol. 150, pp. 467-485. Stewart, T.P.S., (1993). The Influence of Harbour Geometry on the Performance of Oscillating Water Column Wave Power Converters, Ph.D. Thesis, The Queen’s University of Belfast. Whittaker, T.J.T., et al, (1997), The Islay Wave Power Project – An Engineering Perspective, ICE Proc. Water Maritime and Energy, Sept. 1997, No 124, pp 189-201, ISBN: 09650903. Whittaker, T.J.T., et al, (1997), European Wave Energy Pilot Plant on Islay (UK), Commission of the European Communities, DG X11, Contract No. JOU2-CT94-0276.

As expected the grid connection has not been very secure, since June 2001 there have been over 50 events on the grid that have caused the plant to be disconnected. Operational Performance A total of 1800 hours of generating operation have been achieved out of a possible 4000 hours (grid available and plant status ready). Since the plant was commissioned in December 2000, a total of 42000 kWh have been generated. This figure should be considered cautiously; considering the months required to achieve stable operation and the low energy available in the summer 2001, the majority of this power would have been generated during the winter of 2001/2002. New control algorithms are currently being tested in the plant to improve annual productivity.

CONCLUSIONS • •

•

•

The LIMPET project to date is the longest running shoreline OWC device to date. The experience of building the reinforced concrete structure using the methods adopted has stimulated a series of proposed design modifications to permit changes to the manufacturing techniques which will greatly reduce weather dependence and thus lead to a significantly lower installed cost. LIMPET’s control systems have been developed and tested to the stage of continuous stand-alone operation and has logged more hours of automatic operation than any previous devices. This in itself is a considerable achievement. While the power output has been lower than anticipated, the experience gained in determining the reasons therefore will be invaluable for the design of future plant.

REFERENCES Alcorn, R.G., Beattie, W.C., Cully, N., Control Valve Comparison for Oscillating Water Column Wave-power Devices, Proc. 4th European Wave Energy Conf., Aalborg, Denmark, 2000. Boake, C. B., (1998), QUB/Wavegen LIMPET Trials : Preliminary Report, Internal Correspondence. Curran, R., Whittaker,T.J.T., Beattie, W., Raghunathan, S., (1998), Performance Prediction of the Contra-rotating Wells Turbine for Wave Energy Converter Design, Journal of Energy Engineering, American Soc of Civil Engs., USA., Vol. 124, No. 2, Aug 1998, pp 35-53, ISSN 0733-9402.

594