Search

AlternaFuture has been a multidisciplinary research project in Hydrocen to investigate the potential of extreme upgrading of existing hydropower system with a positive effect on the environmental conditions. The project is a desk study and is carried out through developing future scenarios of an extreme upgrading of an existing hydropower system, to create potential for new innovations and solutions from the multidisciplinary scientists within the project. The existing hydropower system in the Mandal river has been applied as a case-study, where the current situation has been the baseline for developing alterative future scenarios and evaluating the economic and ecological results. The project has been divided in three main activities; (A1) mapping the current situation and defining environmental restrictions for reconstruction, (A2) developing physical scenarios consisting of hydropower projects, environmental projects and use of new innovations, and (A3) economic and environmental evaluation of the scenarios. Three main scenarios were developed for the Mandal river; (1) triple installed capacity, (2) maximum flexibility, and (3) flood protection. The hydropower optimization program ProdRisk was used to compare the hydropower operation and water management in present situation with the three scenarios. Energy price forecasts from HydroCen where used to assess the economic income from hydropower production, which in turn where compared with the estimated construction costs of each scenario to consider the economic feasibility. Thereafter, the impacts on the environmental status and recreational value in different parts of the watercourse, including reservoirs, lakes and river reaches are evaluated for three hydropower scenarios and compared with the present situation. In conclusion, it is found possible to realize extreme upgrading of existing hydropower systems, and at the same time in sum have a positive effect on the environmental conditions. It is noted that the positive effects require a significant effort in mapping and planning the environmental measures. For such upgrading consisting of multiple projects, single projects that have severe negative ecological impacts must be cancelled and cannot be included in the final scheme. Planning of such upgrading projects therefore must include environmental experts from the very beginning. The main conclusions from the AlternaFuture research project are presented below. 1. Extreme upgrading of existing hydropower systems can be done while also in sum improving the environmental conditions. 2. Extreme upgrading of hydropower systems that include pumped storage plants are economically feasible if the energy price variability increases sufficiently. For the Mandal river, the necessary increase is in between the 2030 price forecast and the 2030-scaled forecast as described in Memo 1. 3. It is the pumped storage plants that generate the main increase of revenue in the upgrading scenarios. Extreme upgrading without pumped storage plants has not been found economically feasible for any price forecast. 4. The pumped storage plants result in a reduction of the total energy production for the hydropower system, but a higher income. In the current tax regime, the pumped storage plants result in reduced taxation to the local municipalities because some of the Norwegian hydropower taxes are related to produced energy. The taxation to the central government increases depending on the economic profit from the pumped storage plants. 5. There is potential to find new hydropower projects in existing hydropower systems. 6. It is possible to construct a flood power plant to mitigate the flood challenges in the Mandal river. The flood power plant is not found to be economically feasible only from hydropower production, and the remaining costs have to be financed by other stakeholders such as insurance companies or local municipalities. The flood power plant can reduce a 200-year flood to a 20-year flood. 7. The extreme upgrading scenarios have a positive impact on flood mitigation owing to new reservoirs and pumped storage plants. This positive impact has not been quantified in this project. 8. Recommended future work includes developing a best-practice guideline for environmentally friendly upgrading of existing hydropower systems based on the methodology developed and applied in this project. In addition, 18 new research projects have been proposed and are described in Memo 5. © NTNU 2020. The publication may freely be cited with source acknowledgement.

Laboratory measurements using the pressure‐time method showed a velocity or Reynolds number dependent error of the flow estimate. It was suspected that the quasi steady friction formulation of the method was the cause. This was investigated, and it was proved that implementing a model for unsteady friction into the calculations improved the result. This paper presents the process of this investigation, and proposes a new method for treatment of the friction term in the pressure‐time method. Godkänd; 2012; 20121217 (andbra)

Målet med denne masteroppgaven har vært å analysere et billig, robust og enkelt mikro-vannkraftverk bestående av en pumpe-som-turbin (PAT), en egenmagnetisert asynkrongenerator (SEIG) og en asynkrongenerator-kontroller (IGC) koblet i et mikronett, som opererer i øydrift og forsyner en 1-fase landsby-last. Hensikten med denne masteroppgaven har vært å klargjøre det evaluerte systemet for å bli implementert i avsides landsbyer i utviklingsland, som ikke har tilknytning til sentralnettet. Noen av de største utfordringene ved å bruke en asynkrongenerator (IG) alene i frittstående mikronett, er å bestemme størrelsen på magnetiseringskondensatorbankene, samt å sikre stabil spenning og frekvens når landsby-lasten varierer. Målet med IGCen er å sikre en konstant total effekt sett fra generatoren, ved å dumpe overskuddseffekt i en ballast-last. Formålet med dette er å holde spennings- og frekvensvariasjonene innenfor et begrenset intervall. Den nødvendige teorien om de forskjellige komponentene i det evaluerte systemet er presentert. Driftsbetingelsene til det testede systemet ble definert slik at det kunne testes med de komponentene som var tilgjengelige. I tillegg er kravene til el-kvaliteten i mikronettet definert. Først ble IGCen testet i Elektro-maskin-laboratoriet ved Norges Teknisk-Naturvitenskapelige Universitet (NTNU) uten PAT, for å kartlegge oppførselen dens. Deretter ble hele det frittstående systemet testet i Vannkraftlaboratoriet ved NTNU, med PATen som den primære kraftkilden. En simuleringsmodell av det testede systemet er også opprettet i Simscape i Simulink og MATLAB. Formålet er å verifisere laboratorieresultatene. Ved å lage en realistisk men forenklet simuleringsmodell av det testede systemet, kan videre arbeid være å analysere systemet for forskjellige topologier og med forskjellige parametere, som for eksempel ved andre effektfaktorer eller ved å inkludere overføringslinjer. 3-fase IGen som blir brukt i dette systemet konverteres til en 1-fase IG ved å koble magnetiseringskondensatorene i en C-2C kobling. I en slik kobling eksisterer det bare én totallast som sikrer balansert drift av generatoren, for en gitt driftsfrekvens og driftsspenning. Driftsbetingelsene for det testede systemet ble derfor satt hovedsakelig for å oppnå balansert drift av IGen. Både laboratorietestene og simuleringene viser at det testede systemet er i stand til å forsyne en varierende 1-fase resistiv last innenfor de spennings- og frekvenskravene som er definert, ved de definerte driftsforholdene. Kombinasjonen av de komponentene som blir brukt i dette systemet er imidlertid ikke optimal. Testresultatene viser at magnetiseringskondensatorene som er integrert i IGC-skapet er overdimensjonert. Både PATen og SEIGen blir derfor overbelastet. PATen må operere langt fra driftspunktet for beste virkningsgrad (BEP), fordi BEP ved den nødvendige fallhøyden oppnås ved en hastighet som er mye høyere enn driftshastigheten. Oppførselen til PATen er imidlertid bra. Overbelastningen av generatoren resulterer i at linjestrømmene blir målt til å være opp mot 43 % høyere enn den nominelle strømmen for generatoren. Dette resulterte i ekstra effekttap og en betydelig reduksjon i virkningsgraden av generatoren. Overbelastning resulterer også i en lavere forventet levetid for SEIG, noe som harmonierer dårlig med kravet om at systemet må være robust. IGCen klarer ikke å sikre helt konstant 230 V over lasten og dermed klarer den heller ikke å holde en helt konstant totaleffekt sett fra generatoren når landsby-lasten varierer. Når både landsby-lasten og ballast-lasten konsumerer, øker både frekvensen, spenningen, samt den totale effekten, på grunn av karakteristikken til PATen og IGCen. Dette resulterer i et ubalansert system og strømmen i to av generatorlinjene øker ytterligere. Samtidig ble spenningsvariasjonen bare målt til å være 2.6 % høyere enn den definerte operasjons spenningen på 230 V. Det maksimale frekvensavviket ble målt til å være 6.6 % fra den definerte driftsfrekvensen på 50 Hz. Frekvensen ble aldri målt til å være under 50 Hz. Når både ballast-lasten og landsby-lasten konsumerer er det en betydelig andel med harmoniske komponenter i signalene på grunn av kontrollmetoden til IGCen. Den maksimale totale harmonisk forvrengningen (THD) i spenningen og strømmen i landsby-lasten er målt til henholdsvis 4.34 % og 4.33 %. Både laboratorietestene og simuleringene viser at det testede systemet, som består av den forhåndsvalgte PATen, den forhåndsvalgte 3-fase SEIGen og den tilgjengelige IGCen, er i stand til å forsyne en varierende 1-fase resistiv landsby-last innenfor de spennings- og frekvenskravene som er blitt definert for et slikt isolert system, ved de driftsbetingelsene som også har blitt definert. Det blir imidlertid foreslått at størrelsen på magnetiseringskondensatorene integrert i IGCen bør reduseres eller at SEIG bør blir byttet ut, til en annen SEIG med en annen magnetiseringskurve eller høyere nominell strøm enn den som blir brukt i dette systemet. Dette er anbefalt for å redusere overbelastningen av SEIG og bedre virkningsgrad. The aim of this thesis have been to analyze a cheap, robust, and simple micro hydropower plant consisting of a pump-as-turbine (PAT), a self-excited induction generator (SEIG) and an induction generator controller (IGC) connected in an islanded microgrid, supplying a varying single-phase village load. The long-term aim has been to prepare the evaluated system for implementation in decentralized villages in developing countries without connection to the main grid. Some of the main obstacles when using an induction generator (IG), directly connected to an uncontrolled PAT, in stand-alone operations in an islanded microgrid are the difficulties of determining the excitation capacitance requirements, as well as ensuring stable voltage and frequency as the village load connected to the system varies. The objective of the IGC is to ensure a constant total consumed power seen from the generator, by dissipating surplus power to a ballast load, for the purpose of keeping the voltage and frequency deviations within a limited range. The necessary theory of the different components of the evaluated system is presented. The operating conditions of the tested system were defined so that it could be tested with the components which was already selected. Additionally, the power quality requirements of the microgrid were defined, under the assumption that the village load is purely resistive. The IGC was first tested in the Electrical Machine Laboratory at the Norwegian University of Science and Technology (NTNU) without the PAT, to characterize its behavior. Thereafter, the whole stand-alone system was tested in the Waterpower Laboratory at NTNU, with the PAT as the prime mover. A simulation model of the tested system was also created in the Simscape environment in Simulink and MATLAB, for the purpose of verifying the laboratory test results. By creating a realistic, yet simplified simulation model, the system can, as further work, be analyzed for different topologies and with different parameters, for instance for other load power factors or with distribution lines. The three-phase IG used, is converted to a single-phase IG by connecting the excitation capacitors in a C-2C connection. Depending on the operating frequency and the operating voltage, there exists one total load power ensuring a balanced operation of the generator in such a connection. The operating conditions for the tested system were therefore set mainly for obtaining a balanced operation of the IG. Both the laboratory tests and the simulations show that the tested system is able to supply a single-phase resistive load within the voltage and frequency requirements defined for the tested isolated system, at the defined operating conditions. However, the components used are not optimal. The test results show that the excitation capacitors integrated in the IGC cabinet are oversized. Both the PAT and the SEIG were operating in overload. The PAT has to operate far from the best efficiency point (BEP), because the BEP at the required head, is at a speed which is much higher than the operating speed. However, the behavior of the PAT seems good. As a result of the overloading of the generator, the line currents were measured to be up to 43 % higher than the rated current during the laboratory tests. This results in extra power losses and a significant reduction of the efficiency. Overloading also result in a lower expected lifetime of the SEIG, which would contradict the requirement of robustness of the system. Another concern for the tested system is that the IGC does not manage to ensure exact 230 V across the village load, thus it does not manage to ensure constant power seen from the generator, when the village load is varying. Due to the characteristic of the PAT and the IGC, both the frequency, voltage, and total load power increases when both the village load and the ballast load are consuming. This results in an unbalanced system and a further increase in two of the generator line currents. However, the voltage variation is only measured to be 2.6 % from the rated voltage, 230 V. The maximum frequency deviation is measured to be 6.6 % from the rated frequency 50 Hz, but the system frequency is never below 50 Hz. Because of the phase angle control method of the IGC, there are significant harmonics in the signal when both the ballast load and the village load are consuming. The maximal total harmonic distortion (THD) of the village load voltage and village load current is found to be 4.34 % and 4.33 % respectively. This is also within the power quality requirements defined. Both the laboratory tests and the simulations show that the tested system, consisting of the pre-selected PAT, the pre-selected three-phase SEIG and the available IGC, is able to supply a varying single-phase resistive load within the voltage and frequency requirements defined for such isolated system, at the operating conditions defined. However, it is proposed to either reduce the excitation capacitor size in the IGC or that the SEIG in the tested system should be changed to a SEIG with a different magnetizing curve or higher rated current, in order to reduce the overloading of the SEIG and increase the efficiency.

Hydro power plants play an important role in balancing power supply and maintaining a constant grid frequency. With the large water storage capacity, hydro power is in a favourable position with regard to operational flexibility to balance intermittent renewable energy sources, such as wind and solar power. Operation of Francis turbines outside the best efficiency point leads to draft tube pressure pulsations and cavitation, with unwanted effects like noise, vibration and erosion which cause increased wear and maintenance requirements. Flow control is a collection of methods and technologies to modify flow conditions and achieve more favourable situations. Air injection in draft tubes is a common flow control method to attenuate pressure pulsations due to increased compression and reduced wave propagation velocity. These effects can also be favourable with regards to elastic waves in long pipe lines, pressure pulsations and water hammers. This thesis studies the use of flow control methods in hydro power plants, with emphasis on air injection. The presented work is focused around three main experiments. The runner blades at Svorka power plant experienced cavitation erosion, and several measurements were carried out to identify the operational areas with the highest risk of erosive cavitation. By installing plexi-glass windows on the draft tube, we were able to visually correlate the erosion damages and the collapse of cavitation clouds. The effect of a draft tube water injection system on pressure pulsations and cavitation was also studied at Svorka power plant. At installation, the injection system showed positive effects for mitigation of pressure pulsations. However, results from the study presented in this thesis indicate that the injection system has negligible effect on both cavitation and pressure pulsations. This could be due to the amount of air admitted through the runner shaft and variation in turbine submergence because of uncertainties regarding water level of the river and sedimentation in front of the outlet. The effect of air injection was further investigated in a cavitation tunnel experiment, where the coupled effect between hydrofoil cavitation shedding and system dynamics of the tunnel with various gas content was in focus. The results showed a correlation between shedding frequency and the pressure dynamics of the tunnel, but the shedding was not affected by the air content of the tunnel. Nevertheless, such a link might still exist and could possibly be detected by using a different experimental approach. The aim of the last experiment was to assess the transient characteristics of a 1400 m pipe line to evaluate the wave propagation velocity and air content of the pipe. This was carried out by creating water hammers in the system. Furthermore, standing waves were created in the pipes by using a rotating valve apparatus to vary the pipe outlet area and impose periodical variations of flow rate and pressure. A more comprehensive and improved experiment will be planned based on the results and experiences from this study. Collectively, these experiments show that the presence of gas can affect wave propagation velocity and elastic waves, if uniformly distributed. However, the gas can also accumulate in pockets and define a new dynamic element in the system. Gas injection can be used as a flow control method, but it should be taken into account that the gas can collect and create accumulators in the system which can improve or decrease system performance.

There is an increasing need for stable and flexible renewable energy. This calls for better lifetime predictions for Francis runners, which in turn requires more knowledge on the dynamic loads on the runner blades. This thesis outlines an experimental approach to further the understanding of the dynamic loads a Francis runner is subjected to, during off-design operation, transient operation, and during resonance conditions. The objective of the thesis is to provide verification data for numerical analysis, as well as to quantify how the different operating regimes affect the lifetime of a Francis runner, through dynamic loads on the runner. Experiments have been performed on a high head Francis model runner, both for deep part load (DPL), and for part load (PL) with a fully developed Rotating Vortex Rope. The measurements show that the DPL condition causes a large region of back flow, and the PL condition has high amplitudes of the Rehinegan’s frequency. However, these effects are not significant when compared to the pressure amplitude at the runner blade inlet as a result of Rotor-Stator Interactions (RSI). A transient condition was examined, with the turbine undergoing load rejection from the Best Efficiency Point (BEP) to PL, and the RSI amplitude remained dominant at the turbine blades. In order to investigate the RSI phenomena more closely, tests were conducted on a simplified runner blade (a hydrofoil with an asymmetric trailing edge), mounted in a square high-speed channel with angle of attack. The tests show that the damping factor of the hydrofoil increases linearly, with a transition in the lockin region. Particle Image Velocimetry (PIV) measurements showed that the vortex shedding of the hydrofoil exhibited large stream-wise fluctuations in velocity, likely due to wandering of the upper separation point. Further testing on a different hydrofoil geometry reveal that the damping undergoes the same transition through lock-in, even when the vortex shedding amplitude is minimal. Additionally, the tests showed that the two hydrofoils exhibited the same slope in the damping factor change, when plotted against the reduced velocity, or the inverse Strouhal number. Computational Fluid Dynamics (CFD) matches the experimental data, and indicates that the trend continues up to at least 45m=s. A multi-bladed cascade has been tested, for four modes of vibration, and the same transition through lock-in is observed. More interesting, the slope of the damping change coincides with the previously tested hydrofoils, in addition to conforming with the slopes of hydrofoils tested in other works. This suggests that the product of added mass fraction and mode shape integral remains relatively unchanged for fixed-beam hydrofoils. This holds even for blades of differing natural frequencies (one order of magnitude in difference) and for multiple blades with modes coupled through water. The implication of this is that the analysis of a single blade can be extended to a runner, with predictable results.

Friksjonen i massesvingninger i vannkraftverk er ikke tilfredsstillende beskrevet, til tross for at friksjonstapet er viktig for virkningsgraden og stabiliteten til systemet, og optimalisering av kjøreplanen for kraftverket. Forsøk med massesvingninger mellom et lite reservoar og en svingesjakt har blitt gjort. Volumstrøm og trykk ble målt med henholdsvis en elektromagnetisk volumstrømmåler og trykktransdusere. Massesvingningene ble indusert av at en ventil nedstrøms svingesjakten ble stengt. Før ventilen ble stengt, var strømningen gjennom systemet stasjonær med Reynolds tall mellom 26 000 og 52 200 for de ulike forsøkene. Forsøksresultatene ble sammenliknet med simuleringer fra MATLAB. Likningene for transient strømning ble løst med to ulike metoder, en der det ble antatt stivt vann og rør, og en der karakteristikkmetoden ble brukt. For begge løsningsmetodene ble ulike friksjonsmodeller testet. Blant disse var en kvasi-stasjonær friksjonsmodell og en ”oneterm” modell. Noen modifiseringer av modellene ble også undersøkt. Når den kvasi-stasjonære friksjonsmodellen ble brukt i løsningen med karakteristikkmetoden, resulterte det i at svingningene ikke ble dempet godt nok ut. Den gjennomsnittlige relative feilen for de lokale ekstremalpunktene var mellom 4.4% og 6.3% for de ulike volumstrømmene. Med ”one-term” modellen ga simuleringene for mye demping av svingningene for de første fem toppene, og for lite demping for de etterfølgende toppene. Når den kvasi-stasjonære modellen ble brukt i simuleringer der vannet var antatt å være stivt, var resultatene nærmere de målte verdiene enn når karakteristikkmetoden ble brukt. Den gjennomsnittlige relative feilen mellom simuleringene og forsøket var i dette tilfellet mellom 1.45% og 1.70%. Dette var også svært likt de resultatene som ble oppnådd når ”one-term” modellen ble brukt i simuleringene der vannet var antatt å være stivt. En modifikasjon der den kvasi-stasjonære modellen ble kombinert med ”one-term” modellen reduserte de relative feilene ytterligere. Ulike måleteknikker som kan gi mer detaljerte måleresultater ble også diskutert. ”Particle image velocimetry” er foreslått som den beste metoden å gå videre med for nærmere undersøkelser av strømningen ved massesvingninger. The friction in transient mass oscillations in hydropower plants is not well described, despite the importance it has for energy efficiencies, system stability and optimization of the operational schedule of the hydropower plant. Experiments of transient mass oscillations between a small reservoir and a surge shaft have been conducted. Measurements of the flow rate and pressure were done with an electromagnetic flowmeter and pressure transducers, respectively. The oscillations in the hydraulic system were induced by closing a valve downstream of the surge shaft. The initial flow conditions in the experiments were steady flows with Reynolds number between 26 000 and 52 200. The experimental results were compared with simulations in MATLAB. The transient equations were solved with two different solution methods, one where rigid liquid column theory was used and one where the method of characteristics was used. For both solution schemes, different friction models were tested. Among these were the quasi-steady friction model and a one-term model. Some modifications of the friction models were also investigated. When the quasi-steady friction model was used within the method of characteristics scheme, the damping of the oscillations was underpredicted by the simulation compared to the experiment. The mean relative error for the local extrema was between 4.4% and 6.3% for the different flow rates. With the one-term model, the simulations yielded too much damping for the first five peaks, but for the following peaks, it did not yield enough damping. When the quasi-steady model was used within a rigid liquid column scheme, the results were closer to the measured values than with the method of characteristics. The mean relative errors between the simulations and the measurements for this case were between 1.45% and 1.70%. This was similar to the results obtained with the one-term model for the rigid liquid column simulations. One modification where the quasi-steady model and the one-term model were combined reduced the relative errors even further. A few different measurement techniques which will enable more detailed experimental results were also discussed. Particle image velocimetry is suggested as the best method in this case for further investigations of the flow during transient mass oscillations.

As a cheap and available source of renewable energy, regular centrifugal pumps may be run in reverse to act as hydro turbines. Pumps as turbines (PAT) are especially relevant for isolated rural areas, or in developing countries, where efficiency is not necessarily the highest priority. The main challenge, however, is to be able to pick a suitable pump for a given site, which coincides with the available head and discharge. In order to make this decision, one has to be able to predict how a certain pump will perform as a turbine. A lot of work has been done in the past to establish such prediction methods, based on tests and empirical correlations. Many of these methods only predict one optimal point of operation, but as pumps do not have any guide vanes, it will be difficult to constantly operate a PAT at this specific point. Therefore, this may be insufficient. With this in mind, and because of the inaccuracies of these methods, a new approach has been suggested - an approach in which full head-flow characteristics are predicted based solely on the pump's impeller geometry. By assuming symmetry in the velocity diagrams of pump and turbine operation, a method for establishing the PAT characteristic was made. For validation purposes, this method was applied to one specific pump. Experimental results revealed a slight asymmetry, and shortcomings in the initial assumptions. Therefore, the model was modified thereafter, with the introduction of an empirical constant. The predicted characteristic lacks some accuracy in comparison to the real PAT characteristic found through testing, but its optimal point of operation is in line with previous established prediction methods. Still, the modifications, and the empirical constant, may be enhanced further through comprehensive testing of numerous pumps, of various type.

The purpose of this master thesis is to establish and validate a simulation model of the transition from pump to turbine mode of operation for hydropower systems with reversible pump-turbines (RPTs). The model builds on the one-dimensional turbine model presented by Nielsen [Nielsen, 1990. "Transient characteristics of high head Francis turbines". NTH], but uses pump mode of operation as a basis. Simulations have been carried out in MATLAB, and compared with measurements from the RPT rig at the Waterpower Laboratory at NTNU. Simulation of transition from pump mode to turbine mode was conducted by cutting the torque from the motor that drives the RPT in pump mode. This reduces the rotational speed of the runner and it starts rotating in the other direction due to the hydraulic torque from the water, and ends up at runaway speed in turbine mode. The results from the simulations and measurements are fairly similar in pump mode, but very different in turbine mode. At the end point of the simulations the rotational speed is only 1=3 of the speed at the end point of the measurements. There are several reasons for this difference, but insufficient accuracy of the model of the pump-turbine characteristics in turbine mode plays an important role. The model does not take into account that there are two different heads for zero flow depending on whether the RPT is going from pump to pump brake or from turbine to reverse pump mode. The simulation results are also influenced by the assumption of neglecting elasticity and the switch of causality between pump and turbine mode. Torque is the controlling parameter in pump mode, while this is the role of the hydraulic head in turbine mode. All these factors are part of the explanation of why the simulations do not follow the same trajectory as measurements in the H-Q-diagram.

As a cheap and available source of renewable energy, regular centrifugal pumps may be run in reverse to act as hydro turbines. Pumps as turbines (PAT) are especially relevant for isolated rural areas, or in developing countries, where efficiency is not necessarily the highest priority. The main challenge, however, is to be able to pick a suitable pump for a given site, which coincides with the available head and discharge. In order to make this decision, one has to be able to predict how a certain pump will perform as a turbine. A lot of work has been done in the past to establish such prediction methods, based on tests and empirical correlations. Many of these methods only predict one optimal point of operation, but as pumps do not have any guide vanes, it will be difficult to constantly operate a PAT at this specific point. Therefore, this may be insufficient. With this in mind, and because of the inaccuracies of these methods, a new approach has been suggested - an approach in which full head-flow characteristics are predicted based solely on the pump's impeller geometry. By assuming symmetry in the velocity diagrams of pump and turbine operation, a method for establishing the PAT characteristic was made. For validation purposes, this method was applied to one specific pump. Experimental results revealed a slight asymmetry, and shortcomings in the initial assumptions. Therefore, the model was modified thereafter, with the introduction of an empirical constant. The predicted characteristic lacks some accuracy in comparison to the real PAT characteristic found through testing, but its optimal point of operation is in line with previous established prediction methods. Still, the modifications, and the empirical constant, may be enhanced further through comprehensive testing of numerous pumps, of various type.

Denne masteroppgaven er en fortsettelse av prosjektoppgaven "Dynamisk dimensjonering av vannkraftverk". Evanger kraftverk skal rehabiliteres. Kraftverket er driftet av BKK og er Norges 10. største. Ønsket er nå å oppnå mer stabil og hurtigere effektregulering. En måte å oppnå dette er å oppdatere reguleringssystemet. Å skifte til nye injektorer med individuell nålstyring, er en av planene. Masteroppgaven tar først for seg dynamikk i vannkraftverk. Deretter forklares effektregulering i et vannkraftsystem og hvordan kraftmarkedet legger rammer for kraftproduksjonen. Videre vises det hvordan Evanger kraftverk ble modellert i LVTrans i prosjektoppgaven, og hvilke forbedringer som er gjort. Det er tilstrebet å få kraftverksmodellen så lik Evanger kraftverk som mulig. Likevel er informasjonen som er implementert, bestemt ut ifra hva kraftverkselementene i LVTrans etterspør. Drift av kraftverket er simulert, og relevant driftsdata er samlet inn og evaluert. Det er utført frekvensanalyse av alle turbinaggregatene for å evaluere reguleringshurtighet og reguleringsstabilitet. Dette er gjort med regulatorparametere fra BKK, samt regulatorparametere foreslått av LVTrans. Frekvensanalysene gjort i LVTrans er sammenlignet med en frekvensanalyse utført i Evanger kraftverk under idriftsettelsen av en av turbinene. Noen endringer i modellen er gjort underveis for å undersøke samsvar med teoretiske forslag til forbedret effektregulering.

This work has studied the hydraulic and thermodynamics of closed surge tanks for hydropower plants. Surge tanks are applied in hydropower plants to control and reduce the acceleration forces of the water during change of turbine flow. Surge tanks reduce the design pressure of the tunnel and pipes that convey water from the reservoir to the turbines, and enable automatic frequency and power control in the hydropower plant. Closed surge tanks are constructed as underground caverns filled with pressurized air located close to the turbines. The work comprises four different studies. In the first study (1), the benefits and challenges of two different surge tank types, the open and the closed surge tanks, were compared. The state-of-the-art design for these two solutions were described. In the second study (2), a hydraulic scale model of an existing hydropower plant with a closed surge tank was constructed in the scale 1:65 in the Hydraulic laboratory at NTNU. A new method for scaling of the absolute pressure in the closed surge tanks was developed and tested. The results of the hydraulic scale model was compared with field measurements from the prototype, and the accuracy was quantified. In the third study (3) the thermodynamics of the air in the closed surge tank was investigated. The impact of heat transfer from the air to the surrounding media was assessed, and different theoretical models for simulation of the thermodynamics where compared with field measurements from an existing hydropower plant. A new improved theoretical model for the thermodynamics was developed and programmed and tested in the freeware hydropower simulation program LVTrans. In the fourth study (4), the effect of installing a surge tank throttle on governor stability, power control and hydraulic transients was investigated. A new method for evaluation of the throttle effect has been developed, and a quantification of the effects for an example hydropower plant with a closed surge tank was conducted.

Hensikten med oppgaven er å finne en optimal løsning for Sauland kraftverk med tanke på systemdynamikk og stabilitet. Dette innebærer blant annet å sammenligne en luftputeløsning og en svingekammerløsning for hvert av de to fallene. Det vurderes også en struping i svingesjakten i avløpet, på grunn av stort nedsving. Lange vannveier både oppstrøms og nedstrøms turbinen er hovedutfordringen ved utforming. I analysen benyttes analytiske overslagsberegninger, frekvensanalyse og simuleringer av dimensjonerende lastforløp i programmet LVTrans (i LabVIEW), utviklet av Bjørnar Svingen for Sintef. LVTrans utfører numeriske beregninger ved bruk av karakteristikkmetoden. Det er sett spesielt på retardasjonstrykk foran turbinen, reguleringsstabilitet og u-rørsvingninger i systemet. Sauland 1 oppnår bedre reguleringsstabilitet og lavere trykkstøt med luftputekammermodellen. Svingesjaktløsningen er likevel mest aktuell, da den krever mindre tilsyn, er enklere ved inspeksjon og vedlikehold, man har mer erfaring med svingesjakter og systemet har tilfredsstillende stabilitet, så sant svingmassenes treghetsmoment økes noe. Opp til svingesjakten i Sauland 2 er det mer enn to kilometer. Med geometri og turbin som i modellen blir reguleringsstabiliteten dårlig, hovedsakelig på grunn av den lave trykkstøtfrekvens på 0,13 Hz. I luftputemodellen kommer vannspeilet mye nærmere turbinen og reguleringsstabiliteten blir betraktelig bedret med en trykkstøtfrekvens på 0,65 Hz. Det antas som en representativ modell for avløpstunnelen å anta at frispeilarealet øker raskt idet vannoverflaten svinger ned under tunneltaket på 41,45 moh. I så fall blir største nedsving til 40,56 moh. Det viser seg at både nedsvinget og trykket bak turbinen er veldig sensitivt endring i friksjonsfaktor i avløpstunnelen, noe man bør være klar over ved prosjektering og bygging. Simuleringene stemmer bra med kjente analytiske estimater for systemet, både for frekvenser, massesvingninger og forhold i luftputa.

This thesis describes experimental research on the flow inside a vaned diffuser of a centrifugal pump. The measurements were conducted inside one of the ten curved diffuser channels on the single-stage centrifugal pump test rig located at the Water Power Laboratory at the Norwegian University of Science and Technology (NTNU). The measuring technique used during the experiment was laser Doppler velocimetry (LDV), providing velocity data with a high spatial resolution and accuracy. Procedures for operating the test rig and the LDV system are included in the thesis. Detailed quantification of the two-dimensional steady flow is represented as the ensemble-averaged velocity and the root-mean-square (RMS) of the velocity fluctuations at design point and off-design point. Visualization of the flow properties are presented as velocity magnitude plots and velocity vector plots. The results show a well-behaved flow without flow separation at the diffuser walls when operating at design point and off-design point. However, the flow appears to be highly three-dimensional and fluctuating throughout the entire diffuser channel, which is expected in a curved diffuser. The main characteristic of the observed flow is a high velocity core near the concave wall, persisting until the outlet of the diffuser channel. Also a near-stagnation condition of the flow in a low velocity region near the convex wall is observed. The velocity fluctuation distribution has a characteristic S-shape in the upstream part of the diffuser with peak values located at high viscous shear stress zones, while in the downstream part it is uniformly distributed. The results show an overall good repeatability, especially for the downstream part of the diffuser channel, while the data acquired in the upstream part are more scattered. As the purpose of this thesis is to compare the measured velocity data with a Computational Fluid Dynamics (CFD) simulation, the reliability of the data is paramount. Based on the evaluation of the experimental setup and the uncertainty analysis presented in this thesis, the results are concluded to be suited for the intended purpose.

The objective in this Master's Thesis was to verify the stability criteria for a centrifugal pump. Operation of an unstable pump may lead to either exponential or oscillatory unstable behavior of both pressure and volume flow of the water in the conduit system. This is of course unwanted, and fulfilling the stability criteria is desirable. The idea is to test an unstable pump, which was done at the Technical University Berlin (TU) in Germany. Most of the test set-up was already installed, but a new pump and a pressure accumulator was inserted. Unfortunately the experiment was not able to verify the stability criteria systematically, because the set-up was a closed loop without reservoirs; and because the different parameters affecting the stability were difficult to change. The experiment was educational even though the stability criteria were not verified. Different tasks were executed at TU, like installing the pressure accumulator and a pressure sensor, and measuring the pump characteristic. The work with the Master's Thesis also improved the candidate's understanding of the stability criteria and the dynamic behavior of the water in general.A simulation program was made in Matlab, and the aim of the program was to simulate the oscillations of the water in the conduit system. Most of the parameters put into Matlab were measured at TU, except the volume of the air inside the pressure accumulator. This volume affects the frequency of the oscillations a lot, but a good estimation of the volume of the air, made the simulations quite similar to the measured results. In addition to simulating the existing set-up, the simulation program made it easy to vary the parameters affecting the stability criteria. The simulation program was tested this way, and the stability criteria seem to be correct.

Pressure measurements were performed on the Francis runner in the Waterpower laboratory at NTNU. These measurements were divided into start and stop procedures, with the generator on or off. The suggested method that was used to identify the pressure oscillations during the start and stop procedures consisted of finding curves that fit the performed pressure measurements. The MATLAB application Curve Fitting Tool was used to find the curves that fit to the performed pressure measurements. A FFT analysis was then performed on the difference between the measurements and the fit curve to identify the pressure oscillations experienced during the measurements in the laboratory. The FFT analysis showed few signs of interference from the suggested method.The results from the FFT analysis showed that the largest frequency experienced was the blade passing frequency and it was observed during the start and stop procedures, as well as for BEP and part load operation. The guide vane frequency was not observed, this is due to the lack of functioning transducer in the runner blades. The Rheingan frequency had higher peak values during the stop sequence compared to startup. The reason for this is believed to be the chaotic flow pattern experienced in the draft tube cone during the startup sequence. The runner frequency was not observed during the performed procedures. For the occurring frequencies the part load operation resulted in larger pressure amplitudes compared to the start and stop procedures. Elastic oscillations were experienced both upstream and downstream of the turbine. To determine the structural behavior of the runner during the start and stop procedures proper instrumentation in the runner is needed. Improving the instrumentation is strongly advised to further analyze the structural behavior of the runner during the start and stop procedures.By improving the instrumentation on the Francis rig it is possible to perform transient load measurements on the runner during start and stop procedures. There is especially a need to perform measurements along the runner blades to further analyze the effect of the dynamic load on the runner. This can be done by replacing the current defect runner blade transducers, and with a slight modification of the transducer cables to make them stick to epoxy. It is also possible to located more pressure transducers at the upper draft tube cone. This is to perform more measurements on the chaotic flow that was experienced in the draft tube cone during the runner startup. The usage of accelerometers are suggested to determine the occurrence of vibration and cavitation. By installing the accelerometer on top of the guide vane trunnion it is possible to determine the wake and cavitation bubbles occurring from the guide vanes and traveling into the runner. Attaching an accelerometer on the ball-bearings to measure the wear on the ball-bearings while experimenting on how the guide vane angle during startup procedures effects the stress fluctuations on the runner. Strain gauges can also be used simultaneously with pressure transducers along the blade to determine both the pressure and strain occurring along the runner blades. Simultaneous pressure and strain gauge measurements were not performed due to the lack of rapid prototyped runner blades. With the suggested improved instrumentation it is possible to perform these measurements for both the original and rapid prototyped runner blades.

Surveillance of operating power plants is a progressing science in the hydropower sector. The efficiency improvements are closing in on the limits and reducing the costs by improvement of turbine life time seems to be a profitable approach. Continuously updated knowledge of the state of the mechanical parts of the turbine, will improve the ability to plan and execute maintenance work in a more efficient manner and in more suitable periods. It is therefore an important task to address the challenges connected to determination and quantification of relevant parameters, as well as development of reliable and applicable surveillance systems.This study does an experimental investigation of the coherence between hydraulic transients induced by governing and the relevant parameters used to describe the tendency of cavitation in Francis turbines. Theory relevant to these subjects is inquired to describe the problem sufficiently. An embedded application is built in LabVIEW to acquire and process data during experiments. The application is constructed using the Real-Time module of LabVIEW to investigate how this type of acquisition and data processing can be done in stand-alone surveillance systems. Sub-VIs for Real-Time analysis of acquired data are implemented and tested. Amongst other, a modified Gibsons Pressure-Time method is implemented to do transient flow rate measurements in Real-Time. The ability of the system to control external systems are proved by controlling the speed of an electric motor upon user interaction. The experiment results confirm that hydraulic transient from governing in turbines influence the parameters that describe the conditions for inception of cavitation. Real-Time data acquisition and processing prove its applicability to surveillance of these phenomena in operating turbines. No deduction can be made with regard to detection of cavitation with this experimental set up. The experimental set up may further be used to investigate the frequency response of governing in model turbines. The application built in this study may be developed further and more Real-Time processing methods may be implemented to yield more advanced surveillance systems.

Reversible pumpeturbiner blir i økende grad brukt for å balansere kraftproduksjonen. Den økte bruken av reversible pumpeturbiner gjør også at det er nødvendig med bedre kjennskap til hvordan maskinen oppfører seg i uønskede situasjoner. Hvis en pumpeturbin mister kontakten med nettet, vil vannet presse pumpeturbinen i rusing siden generatoren normalt regulerer turbinens hastighet. Dette kan blant annet skje ved en kortslutning eller større feil i nettet.Målingene i denne oppgaven viser hele driftsområdet til en reversibel pumpeturbin. Planlegging og gjennomføring av instrumenteringen har vært en del av dette arbeidet. Målingene har blitt gjennomført stasjonært og dynamisk. De stasjonære målingene har fokus på det ustabile området etter rusingspunktet. De dynamiske målingene ble gjennomført ved å simulere et lastavslag, fra pumpedrift til rusing i turbindrift.Ved sammenligning viser de stasjonære og dynamiske målingene god korrelasjon. Forskjellen mellom målingene kan i hovedsak forklares av den varierende ledeskovlåpningen.De dynamiske målingene viser flere interessante fenomener; den dempede svingningen og det mulige motsatte utslaget. To måleserier viser et stort avvik som kan minne om et trykkstøt. Problemer med instrumentene ved enkelte måleserier, gjør det vanskelig å fastslå om avviket i målingene skyldes et hydraulisk fenomen eller instrumentfeil.

Development of micro-hydropower schemes in rural areas of developing countries is largely depended on simple and affordable systems. The contra-rotating pipe turbine, produced by GreenEnergy is a one-piece and easily operated turbine, designed as the Rolls-Royce of turbines with good operational qualities. Suitable for low head sites, this turbine is easily installed in existing plants, for minimum passage flows or independent schemes supplying electricity to rural areas.The contra-rotating prototype will now be donated to a site in a developing country and the river Una, near Marangu Hotel in Tanzania has been chosen as a potential site. Thereby the turbine could produce clean and needed electricity, and at the same time test its versatility and easy operation.Field work was performed in Tanzania in February, to gain necessary groundwork for this thesis. A proper site review was conducted with a focus on head and distance measurements of the potential scheme. An introduction to the local conditions, both technically and socially was also in focus.The contra-rotating prototype was tested in the Waterpower Laboratory, to assess the condition after the generator upgrades performed by BEVI. Results showed a disappointingly increase of efficiency at design operation. This might imply an inaccurate position of the runner and generator hubs. A full scheme design is presented with all components necessary in addition to an economical review and evaluation local conditions. Designed scheme has a goal of easy operation and minimum maintenance demand, which has been ensured by installation of a Coanda intake and simple control system.Hydrological data of the area has been assessed from data of a nearby area, from a PhD thesis on the hydrological study of the area. This turned out to be one of the large uncertainties presented in this project. The adapted data gave a low expected river discharge, compared to what would secure a sustainable installation.Further progress is now dependent on the partners involved, especially GreenEnergy. Eventual future activities will require thorough discharge measurements from Una.

The flow in Pelton turbines is subsonic, turbulent, multiphase (water, air, and water vapor from cavitation), has high speeds, sharp gradients, free surface and dynamic boundary conditions. A static grid is unsuitable for modeling this mainly due to the turbine wheel and the liquid having a non-stationary relative motion.In recent times, significant progress in CFD simulation has been made, which also is relevant for Pelton turbines.Nevertheless, it is still common to perform costly model tests to test the design of Pelton turbines. There is therefore a need to develop and implement numerical methods that allow for more realistic simulation of flows in a Pelton turbine.In this thesis a meshfree numerical method has been studied, to investigate whether this method can be used to provide a better and more realistic simulation of flows in a Pelton turbine.Smoothed particle hydrodynamics (SPH) is a meshfree numerical method, and has in recent year?s undergone considerable development. The advantage of SPH is that the method is not bound to a lattice and can better manage the free surface of a liquid motion. It uses discrete particles of fixed mass to describe fluid properties, where each particle represents mass and volume. SPH method approximates a function f(x), using a smoothing function W and interpolating between the particles i and j, where the smoothing length h determines the resolution and the number of neighbors that contribute to the properties at a point. There are a number of different interpolation functions.The purpose of this study was to investigate and assess whether the SPH based program DualSPHysics can be a good approach for simulating flows in Pelton turbines.In this paper two test cases relevant for Pelton turbine simulations were performed, a water jet impinging normally on a fixed plate and a simple Pelton bucket geometry. The results were compared with analytical and experimental data. Comparison showed a partially good correlation between the real world and DualSPHysics. In summary, DualSPHysics and SPH emerge as a promising tool in CFD, but this thesis shows that there is some uncertainty concerning the accuracy of the program.

In this project two pump mode tests has been carried out on a model RPT in the Waterpower Laboratory at NTNU. The measurement set up and execution of the tests were the same, except for the method of dissipating energy in the system. In the first test one of the feed pumps in the basement was used as energy dissipator. In the second test a throttle valve was used as energy dissipator. The objective has been to see how the Waterpower Laboratory is suited for such a test. It was also important to test the two different dissipation methods and evaluate which should be preferred when performing a pump mode test. During the tests it was observed that the guide vane angle kept changing. When the results were processed it became evident that the variations in guide vane angle during the tests had a big influence on the measured pump curves. Variations in guide vane angle has not previously been observed during turbine mode testing. It is presumed that the variation in guide vane angles is due to play in the guide vane system and the design of the guide vanes. The guide vanes are Francis vanes, and not RPT vanes. It is assumed that it is the sharp trailing edge of the guide vanes that causes turbulence over the vanes in pump mode, thus making the guide vanes move about within the play of the guide vane system. It is suggested that the guide vane system is replaced with an RPT guide vane system for further testing in pump mode. The two dissipation methods tested were both effective. The throttle valve works best for achieving 0 flow, while the feed pump is easier to regulate. None of the methods showed signs of high noise or vibrations, and both may be used for future tests.

During the spring of 2008, Jón Bergmann Heimisson developed a pump design program in Matlab. The program has been further developed during the work with this thesis, as well as in the author's preceding project thesis, giving key information for an existing pump design. The aim of this Master's thesis has been to verify the calculation of pump characteristics and velocity profiles at the impeller outlet through testing.A detailed description of the relevant theory regarding pump design has been presented, and different calculation models for the pump characteristics have been examined. The analytical approaches for calculating the performance data have been implemented into Matlab, and a comparison of the different calculation models has been performed. A multistage centrifugal pump has been used for verifying the velocity profiles, and the pump characteristics have been compared to the different calculation models presented in Matlab. Measurements of the velocity profiles were carried out in Typhonix' laboratories at Varhaug using a pitot-static probe.The results achieved from the comparison of the characteristic curves calculated in Matlab showed that the models provide quite different results. Some of the methods widely used in the literature proved to deviate significantly from the measured results, while other and more advanced methods provided better results.The results achieved from testing the velocity profiles with the pitot-static probe were not as good as desired. The measured velocities and flow angles did not correlate well with the analytical solutions, and the results are partly unreliable. Some of the trends regarding changes due to increased volume flow or rotational speed were found, but the exact values could not be trusted. The pitot-static probe is an intrusive method, and it probably disturbed the flow in a way making good results difficult to achieve.

Both previous experiments and CFD simulations show that bends have a significant influence on the downstream flow field, especially by causing so-called ?skewed velocity profiles?. Based on the simulations carried out in OpenFOAM during this thesis, the axial velocity profiles downstream of bends of a few selected geometries have been plotted and described in more detail than what the author has managed to find in previous work. The findings were divided into characteristics of velocity profiles in ?Plane AA? and ?Plane BB?, which are defined in Figure 2.1, while the bend simulations were carried out on four different bend geometries of two different angles and two different relative radiuses, which are listed in Table 4.1.In general, it was found that a sharper bend angle and a smaller relative radius lead to more skewness in the velocity profiles. Bends with small relative radiuses also tend to have sharper convective velocity gradients, i.e. sharper change in velocity over change in position, especially shortly after the outlet of the bend.In the case of the 45 degree bend (relatively small bend angle), the influence of using various relative radiuses (e.g. R/r=2 vs. R/r=8) seemed to make less impact on the velocity field skewness than that of the 90 degree bend. A number of simulations were carried out on a mesh of the NTNU Tokke spiral casing model, using ANSYS for meshing and CFX for simulation and post-processing. Radial velocity profiles at the outlet of the stay vanes were plotted against the angular position at the outlet. As expected due to the effect of the14stay vanes on the velocity distribution, the velocity field appeared divided into 14 velocity profile ?peaks?(see Figure 5.4.3.2a and b), each corresponding to one of the 14 stay vane channels that the water passes through on its way towards the guide vanes, and eventually into the runner.Based on the plots of the radial velocity components, it appears that the general shape of each velocity profile looks like a single, skewed and ?fang-shaped? peak. The fang-shape of each profile iscaused by a skewness leaning towards the inner curveof the channel curvature, where the gradient of the radial velocity over angular location (𝜕𝑢/𝜕𝜃) is larger at the inner part of the channel than at the outer part. This phenomenon is very similar to the skewed velocity profile in Plane AA that occurs shortly after a flow field enters the inlet of a bend, before it switches and starts skewing towards the side corresponding to the outer curve of the bend curvature.Although the spiral casing and stay vane design of the Tokke spiral casing model successfully achieves similar radial velocity profile shapes from each channel, the magnitude of the velocities, and thus the volumetric flow, going through each channel, varies more than what should be optimal. E.g., the peak velocities going through the first and the last channel. which are the channels with the lowest and highest peak velocity, respectively, have an absolute velocity difference of about 25%.

The aim if this Master?s thesis was to establish relation between pressure pulsation amplitude in a model and a geometrical similar prototype. Model test has been conducted in the Water Power Laboratory at Norwegian University of Science and Technology (NTNU). The data collected from the model test has been compared with corresponding data from field measurements at the geometrical similar prototype. Three operation points have been compared; these were 42 %, 50 % and 75 % load. The comparisons were made from three different locations; draft tube cone, runner blade and vaneless space. As a scale-up relation the A/H fraction has been tested, but none of the pressure transducer showed a tendency to follow this relation. Moreover it have been conducted pressure pulsation measurements on the model turbine for different sigma levels. These measurements were conducted at 50 % load with four different sigma levels with the range of 0.026 to 0.050. The sigma variation influence proved to be significant in the draft tube, and marginal in the runner. For the vaneless space the influence could be neglected. However the results from all the transducers showed a tendency of an increase in pressure pulsation amplitude with decrease in sigma level.

Denne masteroppgaven omhandler en dynamisk fluidanalyse av et system bestående av to i utgangspunktet konsentriske sylindre som kan bevege seg i forhold til hverandre. Oppgaven har tatt for seg utviklingen av en dynamisk mesh modell som brukes som grunnlag for den dynamiske bevegelsen. Masteroppgaven bygger videre på arbeidet som ble gjort i prosjektoppgaven "Flow in cylindrical oil damper" der det samme problemet ble løst med en kvasi-statisk metode. Modellen som har blitt utviklet baserer seg på en akselerasjonsrespons som er transformert slik at den kan brukes på en dynamisk mesh løser. Modellen tar inn variabler som amplitude, fasevinkel og frekvens og den er basert på en sinusbasert akselerasjonsrespons. Den dynamiske modellen har blitt testet mot et eksempeltilfelle. Testtilfellet var et hydraulisk stempel og ingen synlige avvik ble oppdaget under testing med den modellen. Det matematiske grunnlaget for dette systemet har blitt utvidet enda mer i masteroppgeven og kombinert med det som ble definert i prosjektoppgaven dekker det mesteparten av den kjente teorien om dette systemet. Videre så har den dynamiske modellen blitt testet med forskjellige viskositeter, geometrier og frekvenser i den hensikt å kartlegge et dimensjonsløst tall som kunne bli brukt til å skalere systemet. Flere forskjellige simuleringstilfeller som geometri, frekvens, amplitude og viskositet har blitt undersøkt og mesteparten av disse variablene har blitt krysssjekket for å kartlegge hvordan de påvirker systemet. De forskjellige variablene for systemet har blitt kartlagt og oppsummert for de undersøkte tilfellene som har vært analysert i denne oppgaven. Trendlinjene fra disse resultatene kan brukes for forskjellige geometrier, så lenge de dimensjonsløse tallene er innenfor de samme områdene som i denne undersøkelsen. Alle resultatene fra simuleringene har blitt testet opp mot det som finnes av aktuell teori og det er ikke funnet noen ting som antyder at denne modellen ikke fungerer for dette formålet.

The high concentration of sediments in the rivers near the Andes and the Himmalayas, is a large problems for run-of-river power plants. Small particles of quartz, which is a hard and sharp mineral, damage the turbine and causes reduced performance, shut downs of power plants and often extensive repairs. At Cahua power plant in Peru, turbine components need to be changed annually since as much as 50 kg/s of sediments passes through the turbines. Together with SN Power, Dynavec have installed a bolted turbine with a chemical coating, which desirable will reduce the maintenance interval.This master thesis has the objective of predicting sediment erosion in Francis turbines. By the use of CFD analysis of the stay, guide and runner vanes, the erosion areas and intensity can be found. This is investigated by implementing particles in the numerical simulationsand utilizing Langrangian particle tracking to predict the particles path through the turbine. With a numerical erosion model, the erosion areas and degree of intensity can be determined. Results show that the numerical simulations give reasonably prediction of erosion trends.Vulnerable areas are detected at the upper and lower covers and runner blade surfaces, which agrees with the observations at Cahua power plant. A comparison between the numerical and theoretical models for erosion on material is made with a verification model. The model is a high velocity jet test rig which test sediment influence on materials. The simulations indicate similar erosion trends, however, the erosion intensity rate does not agree with results from experiments.Sediment erosion is a highly complicated phenomena, and is dependent upon several factors. Numerical erosion models, will, with some modifications, give a good prediction of erosion areas and relative values of the erosion intensity. The simulations can be utilizedto predict the location of severe erosion and to estimate maintenance intervals.

Det har lenge vært ønskelig å bygge en kavitasjonsrigg ved Vannkraftlaben for å forske på modulert kavitasjon på vingeprofiler. Formålet med oppgaven er å prosjektere en slik rigg, samt beskrive nødvendig instrumentering og utarbeide et prisoverslag.Det ble først hentet inn bakgrunnstoff fra andre kavitasjonsrigger rundt om i verden for å ha et grunnlag for hvordan riggen på Vannkraftlaben skulle utformes. Etter at dimensjonerende størrelser som volumstrøm og størrelse på testseksjon ble fastslått, er utregning av størrelser som høyde og lengde på hele riggen bestemt. Komplett opptegning i 3D av en ideell kavitasjonsrigg ble deretter laget.Etter at to alternativer for plassering av riggen i laben ble undersøkt, ble det bestemt at alternativ nummer to var det beste. Ved alternativ to skal pumpen og motoren til kavitasjonsriggen stå foran trykktanken til Francis-riggen. Denne plasseringen var mest hensiktsmessig med tanke på plass i laben, og vil gi mist konflikter i fremtiden med tanke på en eventuell utbygging. Plasseringen var også litt mer hensiktsmessig med tanke på rørlegging opp til galleriet, da man unngår konflikt med traverskrana som strekker seg over store deler av laben, med unntak av et par meter nede ved kortveggen mot sør. En ny 3D-modell av riggen tilpasset Vannkraftlaben ble så tegnet i Inventor. Modellen avviker fra ideell geometri grunnet punktene nevnt ovenfor.En vurdering av nødvendig instrumentering er blitt presentert og diskutert. Det er blitt konkludert med at en PC, flowmeter, trykksensorer, oksygenmåler, høyhastig-hetskamera og PIV er nødvendig i kavitasjonsriggen.Til slutt ble det utarbeidet et prisoverslag for kavitasjonsriggen. Total pris er estimert i underkant av NOK 2 500 000. I denne prisen inngår ikke et høyhastighetskamera da dette allerede er tilgjengelig på Vannkraftlaben.

Det er i denne oppgåva sett på mulegheitene for å gjennomføra eit storskala u-røyrforsøk ved Vannkraftlaboratoriet ved NTNU. Eit slikt forsøk må kunna måla transient hastigheitsprofil til vatnet som oscillerer. Det er foreslått å bruka PIV (Particle Image Velocimetry) til måla dette. I tillegg må vassnivået i u-røyret overvakast og loggast for å visa korleis svingingane dempar seg ut. Det er ønskjeleg å finna ein samanheng mellom korleis hastigheitsprofilet ser ut, og utviklar seg, og dempinga. I tillegg er det interessant å undersøka korleis ulike svingefrekvensar og startamplitudar påverkar dempinga. Ei muleg løysing på oppsett og målemetode er funne, og bør kunna gjennomførast i Vannkraftlaboratoriet.I tillegg er det sett på to ulike dempemodellar for u-røyrsvingingar som er samanlikna med tidlegare målingar gjort på eit mindreskala u-røyr. Begge desse dempemodellane viser seg å gi for lite demping for større amplitudar. Det meste av teori og forsøk som er gjort på demping av u-røyrsvingingar dreiar seg om småskala oppsett med låge staramplitudar. Strøyminga får dermed relativ lågt Reynoldstal samanlikna med eit fullskala kraftverksystem. Det er usikkert om slike dempemodellar kan overførast til fullskala vasskraftsystem.

Whilst growing demand for energy in Nepal and neighbouring- countries, investments in hydro- power projects appears continuously, both by local and multinational companies. As a consequence of this, Turbine Testing Lab was founded in 2011 at Kathmandu University in Nepal. The laboratory is rapidly developing, and is currently implementing a Francis turbine test rig. A long- term goal for the laboratory, is to execute model tests according to IEC 60193, which is the standard used in model tests of hydraulic runners.When determining the hydraulic efficiency of a runner, a central parameter is the discharge measurement. According to IEC, any secondary device used to measure the discharge shall be calibrated in sitù against one of the following primary methods: the weighing method, volumetric method or the moving- screen method.The aim of this work has been to develop and design a primary method for calibration of the flowmeter at TTL. Based on an evaluation of the mentioned primary methods, as well as economy, accuracy and correspondence with technical staff, a volumetric method is chosen as calibration principle. The principle of the method is based on collecting water into a tank with a known geometry, and by execution of level- and time measurements calculating the discharge.The calibration rig consists of a flowmeter, inlet nozzle, deflector mechanism, emergency weir, calibration tank, level measurement, time measurement, drainage system, emergency weir, and an upgraded pipe run at the measuring section of the flowmeter.It is developed a LabView program, for logging and processing the voltage output from the flowmeter. The other measurements executed is registered manually in a separate calibration sheet for data processing. An evaluation of the accuracy in the calibration method is conducted, which may be used to determine the total uncertainty when the calibration facility is installed. Provided design and installation according to ISO 8316, the accuracy in discharge measurement with the volumetric method lies within $\pm$ 0,1- 0,2 \%. The major uncertainty contribution in the method lies in the determination of the volume collected, and the corresponding tank calibration.It is proposed to reduce the diameter of the flowmeter from original size of 400 mm to 250 mm. Reduction in diameter and upgraded pipe run will increase accuracy of the flow measurement, a result of improved flow pattern at the measuring section. When the primary calibration rig is installed at TTL, this will be a large step towards an IEC- approved laboratory in Nepal, which open doors against an international market.

Model tests on a high head francis turbine has been conducted at the waterpower laboratory at NTNU. Simultaneous measurements of the pressure were performed on different parts of the waterway. Compact RIO hardware and Labview software from National Instruments was used for data aquisition. Pressure sensors were placed at the inlet, vaneless space, runner channel, draft tube cone and draft tube outlet in the rig. Generator torque and the rotational speed of the runner was also measured. Spectral analysis was performed on all measurements to map out frequencies and corresponding amplitues at various load conditions. The RSI induced pulsations dominates the vaneless space and runner channels. Pressure pulsations are dampened across the runner and sensors at the trailing edge display values that are significantly larger than sensors positioned at the outlet. RSI pulsations are dependent on the flow and relative amplitudes in the vaneless space is reduced with over 90% when the load is lowered from 100% to 50%. Vortex rope frequency is detected in the runner channels and the generator torque. This suggests that the low frequency oscillations causes a mass oscillation through the system that affects the output of the generator. An air lekeage through the shaft was detected during testing. The incoming air resulted in an excessive draft tube vortex, thus the results obtained in this work are not directly comparable to field test conditions.

Europe's energy production is experiencing a shift towards larger volumes of renewable energy. This developement, however benificiary, poses several challenges. One of them being the lack of regulation, as the energy is available when nature permits. As a consequence, Norwegian hydroelectric plants, risk operating under conditions not anticipated in their planning/construction stage. This occurs during large influx of unregulated power, where grid stability needs to be maintained by these plants. Presented herein is a model implemented to investigate grid influence on system components, down to the waterway. The model was verified on the Driva hydropower system. The background data was provided by Norconsult, as the measurements could not be conducted be the author himself. This was due to circumstances around the plant owners.Simulated load rejections overestimated the runaway speed and slightly underestimated the pressure surge, compared to measured data. Simulated behaviour of remaining online unit seemed reasonable and maintained good stability. The model output compares well with analytical solutions. The turbine model behaved as expected during transient load changes, however unit output did not change as expected when changing the grid frequency. This was a result of the governor models not operating as expected. The model was also able to simulate highly undesirable conditions in the wateray due to grid frequency fluctuations.The program generally compare well with rejection trial data as well as expected physical behaviour of the various components. A few points for improvement was suggested, including a further investigation into the governor models.

Hydro power plants are operated in a different manner than what they used to be, due to increased focus on economy, and less on operating on best efficiency point. This creates new challenges related to wear and tear of the plant. In order to maintain and avoid degradation, the need for control is increased. Installing sensors in a hydro power plant may be difficult, and modeling and estimating parameters could be a solution. This thesis will cover real time modeling of flow systems, with particular focus on the Kalman filter. The filter is an important part of control engineering, but the utilization in hydro power seems to be limited. The goal of this thesis is to understand how the Kalman filter works for hydro power applications, and how it can be implemented in LabVIEW. The Kalman filter investigated is the nonlinear Discrete Extended Kalman filter. The case chosen is the estimation of flow, based on two different pressure losses. The Kalman filter program was run at different operating points in order to investigate the filters function on the dynamics of the system. The experimental rig used was the existing Swirl rig at the Water Power lab at NTNU. Some modifications had to be made in order for the rig to fit the experiment specifications. One of the main valves were changed, and some extra pressure outlets were made. Both flowmeters, absolute pressure transducers and differential pressure transducers were used for the experiment, all of these calibrated by the calibration program presented. The calibration includes uncertainty analysis. Both the calibration program and the Kalman filter program is presented step by step in order to describe the features and logic behind the programming. The main part of the results seemed to coincide with the Kalman filter theory. The estimations of the flow based on pressure loss over the valve seemed to follow the measured values, but the estimations for the pressure loss over the swirl generator did not. Some of the estimations showed reduction in loss compared to the measurements. These and more results are presented and discussed.

Det er en pågående diskusjon innen vannkraftmiljøet angående hvorvidt trykkpulsasjoner målt i modellturbiner kan skaleres og brukes til å forutse hvor store trykkpulsasjonene blir i prototypturbinen. I senere tid har det oppstått flere problemer knyttet til trykkpulsasjoner i Francisturbiner. Det er antatt at dette har ført til blant annet maskinhavari. Det er derfor viktig og interessant å sammenlikne måledata fra eksisterende modeller og korrespondernde prototypturbiner. Ambisjonen for denne masteroppgaven var å samle inn slik informasjon fra ulike kraftverk, men av forskjellige grunner ble det bare mulighet til å sammenlikne trykkpulsasjoner fra ett kraftverk. I denne masteroppgaven har jeg fått mye hjelp fra næringslivet. Rainpower, Norconsult og Statkraft har bidratt med tilgang på måledata fra både modell- og prototypturbin. Så store mengder data behandles best ved hjelp av MATLAB. Ifølge IEC er det mest hensiktsmessig å benytte Fast Fourier transformasjon når en analyserer trykkpulsasjoner i Francisturbiner. Det er derfor valgt å benytte denne metoden i denne masteroppgaven også. Det er utført en dimensjonsanalyse for å danne en teoretisk beskrivelse av hva trykkpulsasjoner i Francisturbiner avhenger av. Med andre ord hvilke dimensjonsløse parametere som ''styrer'' trykkpulsasjonene. Selv om denne masteroppgaven bygger på målinger fra ett kraftverk er resultatene interessante. I motsetning til hva som var tidligere antatt, virker det som om skalerte trykkpulsasjoner generelt blir større i prototypturbiner enn i modellturbiner. Videre har det vist seg å være vanskelig å finne en måte å skalere trykkpulsasjoner målt i modellturbiner slik at de samsvarer med de som forkommer i prototypturbinen. Forhåpentligvis er dette starten på innsamling og sammenlikning av trykkpulsasjonsmålinger i modell- og prototyp Francisturbiner. Mye kan læres og erfares ved slikt arbeid. Det er gitt begrunnelse og forklaring av den metoden som er brukt for å analysere måledataen i denne masteroppgaven. Håpet er at det videre arbeidet skal kunne bygge på denne masteroppgaven og bruke samme metode. Dersom det er av interesse, bør det være et langsiktig mål å danne en standarisert metode for å måle og sammenlikne trykkpulsasjoner i både modell- og prototypturbiner.

Det er i denne rapporten gjennomført dynamisk analyse av to kraftverk under prosjektering av BKK, Aldal Kraftverk og Nye Frøland Kraftverk. Analysen omfatter elastisk simulering av opp- og nedsving i svingesjakter og retardasjonstrykk ved avslag eller pådrag, samt frekvensanalyse for reguleringsstabilitet. For simulering er det utviklet et generelt program som benytter karakteristikkmetoden for å beregne retardasjonstrykk og massesvingninger i systemet, mens et program utviklet av en annen masterstudent er brukt under frekvensanalysen. Mye av fokuset i dette arbeidet er å utvikle et simuleringsprogram for beregning av massesvingninger og retardasjonstrykk som er så generelt som mulig og dermed kan benyttes for andre kraftverk enn de to som er analysert her. Kildekoden til dette programmet er å betrakte som åpen, og kan fritt benyttes av andre for egne simuleringer eller videreutvikling.Simuleringer for Nye Frøland kraftverk har vist at foreslått svingesjakt er underdimensjonert, og overflom vil inntreffe ved pådrag med etterfølgende avslag i mest ugunstige fase. I tillegg skaper et trangt tverrsnitt opp til svingesjakten problemer med trykkstøt og regulering. Videre simuleringer viser at om dette tverrsnittet utvides med en meter i diameter resulterer det i akseptable trykkstøt og en rask regulering. Det konkluderes derfor at svingekammer må utvides, samt flyttes lengre opp i terrenget for å unngå overflom. Svingesjakt i Aldal Kraftverk er foreslått plassert i forbindelse med svingesjakt til Aldalselva, som er det nederste av to bekkeinntak i dette kraftverket. Simuleringer viser at svingesjakten er dimensjonert tilstrekkelig for å unngå overflom, men nedsvinget vil nå ned til tilføringstunnelen for bekkeinntaket. Dette er ikke kritisk for kraftverket, men vil føre til økt sedimenttransport. Det er også konkludert med at antagelsen om at bekkeinntak mot Kvernelva kan neglisjeres ikke stemmer. Bekkeinntakets lengde er i den størrelsesorden at oppsving til bekkeinntaket vil finne sted, med mindre det installeres en enveis ventil i nevnte bekkeinntak. Det konkluderes derfor med at svingesjakten bør flyttes opp til knutepunkt mellom øvre bekkeinntak og driftstunnel. Forslag om et nytt svingesjaktprinsipp er lagt frem, da opprinnelig prinsipp får en lengde på sjakt i størrelsesorden med tilføringstunnelens lengde. Det har i alternativet nevnt ovenfor vist seg at en god regulering ikke kan oppnås. Det anbefales derfor å vurdere å innføre en ekstra svingesjakt nærmere turbinen for å bedre reguleringen.

Start and stop procedures affect pressure oscillations throughout a hydropower plant. A desire to study how pressure oscillations behave during these dynamic conditions was the basis of this report. Instrumentation, experimentation and measurement analysis was conducted on a Francis model turbine in the Waterpower Laboratory at NTNU. Eight pressure transducers were calibrated and used during the experiments. Two transducers were installed in the draft tube below the turbine. One was placed in the vaneless space between the guide vanes and the impeller vanes. Three pressure transducers on an impeller vane and two transducers located at the inlet were also included in the experiments. Frequency analysis (PSD) was carried out for all the measurements to explore various pressure oscillations. Except for the low frequent oscillations (< 30 Hz), definite frequencies repeatedly dominated the frequency domain during start/stop as well as for steady state operation. The impeller vane oscillation showed an increase in pressure amplitude during guide vane closing. A bigger amplitude increase was registered for BEP than for part load and full load operation. The guide vane frequency was located in and only in the runner. The amplitude of the guide vane frequency was significant and was located for all studied operational points. The power of this oscillation decreased during guide vane closing. One specific frequency arose the question of an overtone phenomenon for the water hammer oscillation, a phenomenon, were the fundamental frequency is three times higher than the customary water hammer frequency.

The goal with this master thesis was to establish Hill diagrams and improve a Kaplan turbine intended for use in Afghanistan. The turbine efficiency has been tested in setting 1 and 2. Turbine efficiency in setting 3 and 4 could not be tested because the runner blades interfere with the housing making it impossible to rotate the turbine. The efficiency was tested with an effective pressure head ranging from 2 to 8 meters. Best efficiency point was not reached because of limitations in the test rig making it impossible to reach a lower effective head. The best efficiencies tested in the two different settings are presented in the table below together with the uncertainty in the actual test point. All tests are done according to the IEC standard for model testing of hydraulic turbines. The computational fluid dynamics (CFD) simulations done on the inlet bend indicates that the bend should be rounded and flow controllers should be extended over the entire bend. This should be considered to get a more even velocity distribution at the inlet of the guide vane. An alternative placement of the lower bearing was designed but is discarded because of the disadvantages the modification leads to. High wear due to sand erosion on the seals causing high maintenance and costly stops makes the solution not optimal for use in water with high sand content. The runner blade design is checked against the design procedure presented by Professor Hermod Brekke in Pumper og Turbiner and found to be satisfying. It is concluded that time should rather be spent on optimizing the inlet of the turbine. Fluctuations in the measurements make it necessary to change the measuring equipment or search for error in the existing equipment before further tests can be carried out. In order to be able to test in setting 3 and 4 the runner needs to be placed while the blades are fixed in setting 4.

Over the last few years Norway has seen an increasing number of hours where the grid frequency exceeds the required limits (49.9-50.1Hz). To improve this situation one alternative is to implement hydropower governing with quicker response time. However, long conduits and oscillatory flow set strict requirements to the hydropower system stability and turbo set governing. This thesis establishes a simulation program based on the structure matrix method for stability analysis of hydropower systems. The method is implemented in a Matlab program to study the oscillatory flow in the frequency domain. Implementation of frictional influence, turbine characteristics, and alternative governing has been given special attention. The program is validated through comparison with measurements and previous analysis at Kongsvinger and Tafjord power plants. The program simulations generally compare well with physical dynamics of the two systems. Further a stability analysis of speed governing at Aldal power plant has been performed. Finally some alternative control systems are discussed.

Hovedfokus for oppgaven er å analysere nytten av en pumpe som et tillegg til kraftverket som ble vurdert i prosjektoppgaven Alternative tekniske løsninger for utbygging av lavtrykks kraftverk . Kraftverket er plassert ved overføringstunnelen mellom hovedmagasinet og inntaksmagasinet til Tyin kraftverk. Prosjektoppgaven anbefalte å installere en Kaplan-turbin med kapasitet på ca. 9,5 MW og 35 m3/s i kraftverket. Dette er derfor grunnlaget for denne oppgaven.Hovedbegrunnelsen for en pumpe i systemet er at mye av tilsiget går direkte i inntaks-magasinet (Torolmen), som har relativt liten magasineringskapasitet. En må derfor noen ganger kjøre Tyin kraftverk selv om prisene er ugunstige og en heller skulle ønske å magasinere vannet. Ved å pumpe vannet fra for eksempel deler av vårflommen rundt 30 meter fra Torolmen og opp i det store magasinet i Tyinsjøen, kan en unngå produksjon i Tyin kraftverk til de mest ugunstige prisene. Kostnadene til investering i og drift av pumpa veies opp mot verdiøkningen som oppstår når dette vannet kjøres både gjennom Torolmen kraftverk og Tyin kraftverk etter at prisene har økt. Verdiøkningen gjelder hovedsaklig for produksjonen i Tyin kraftverk siden fallhøyden her er over 30 ganger så stor som pumpehøyden.Fem forskjellige pumpestørrelser er vurdert, og brukt i simuleringer av mengde pumpet vann og inntekt fra prisdifferansene i årene 2006 til 2011. Det er også gjort overslagsberegninger på hvilken prisdifferanse som er nødvendig for å kunne betale ned kapitalkostnaden for pumpa med ulike mengder pumpet vann. Resultatene sannsynliggjør lønnsomheten for pumpa. Pumpa gir økt fleksibilitet for hele systemet til en forholdsvis liten kostnad. Den prisdifferansen som skal til for å tjene inn pumpa er liten. I et marked som periodevis vil ha stort overskudd med stor sannsynlighet for kortere perioder med priskollaps, kan en pumpe som tar unna vannet være meget lønnsom. Stabiliteten til kraftverket er også vurdert. For å unngå for stor treghet i systemet og store trykkstøt som kan skade anlegget er en nødt til å ha en svingesjakt i systemet. Denne bør ha et vannspeilareal på minst 45 m2 for å unngå ustabile U-rørssvingninger mellom sjakt og magasin. Ved avslag etter pådrag får en også mye vann som må tas vare på. For å unngå overløp, må en finne en måte å ta unna dette vannet på, for eksempel øke tverrsnittet av sjakten ytterligere. Trykkøkningen foran turbinen blir på over 30 prosent, selv med et vannspeilareal i sjakten på 70 m2. Det trengs derfor en ventil eller luke som åpner seg ved brått avslag for å unngå dette. Tidskonstantene for vannveien gir en indikasjon på at systemet blir tungregulert og tregt, og et kraftverk ved Torolmen bør kjøres på et sterkt nett. Det vil mest sannsynlig si at kraftverket må kjøres samtidig med Tyin kraftverk. For å minske belastningen på nettet og sikre at pumpa ikke får fullt moment så raskt at det skader akslingen og motoren, kan en installere en soft-starter eller pumpe med turtallsregulering.

I denne oppgaven har en Kaplan turbin produsert i Afghanistan blitt undersøkt. En fullstendig virkningsgradanalyse er gjort i laboratoriet, på to av fire løpehjulsvinkler. Like før innløpet til turbinen er det et 90 grader bend, som har blitt simulert i Ansys Fluent. Forbedringer av bendet er foreslått. Turbinen er laget av Remote HydroLight for bruk i Afghanistan. Målet med turbinen er at oppbygningen er enkel nok til at innbyggerne kan produsere og vedlikeholde turbinen på egenhånd. Det betyr at turbinen må være modifisert i forhold til en tradisjonell Kaplan turbin. Sagt med andre ord, er et enkelt design viktigere enn en høy virkningsgrad. Forenklingene av turbindesignet tatt i betraktning, er beste virkningsgrad høy; 85.32 % for løpehjulsposisjon 1, og 87.75 % for posisjon 2, med en usikkerhet på 0.5 %. Reduserte parametere er brukt. Dersom fallhøyden økes, ser man en liten økning i virkningsgraden. Dette kan skyldes lavere friksjonstap ved høyere Reynolds tall. For posisjon 2 ser man den motsatte effekten; virkningsgraden minker litt med høyere fallhøyde, for høy volumstrøm. Dette kan være fordi høyere volumstrøm gir høyere tap i bendet, fordi strømningsprofilen blir mer ujevn. Løpehjulskovlene har fire ulike posisjoner, mens ledeskovlene ikke kan justeres. Posisjonene er markert med små hakk på bladene, og å finne eksakt samme posisjon når skovlene har blitt flyttet på er vanskelig. Målinger gjort på samme skovlåpning gir derfor varierende resultat. Bendet har to strømningsrettere og effekten av disse har blitt simulert. Simuleringene viser at den nedre strømningsretteren har en stor positiv effekt på strømningen i forhold til å ikke ha strømningsrettere. Den øvre strømningsretteren viser liten eller ingen effekt på strømningen og kan derfor bli fjernet. Dersom den nedre strømningsretteren flyttes mot høyre, er den positive effekten på strømningen enda større, ved at den akselererer strømningen i den indre delen av bendet. Simuleringer har blitt sammenlignet med Pitot-målinger gjort i laboratoriet, og de viser de samme tendensene i strømningen.

A synthetic jet is a flow control device which injects a pulsating jet flow of high-momentum fluid into a boundary layer near the wall confining a main flow. The technique is used to prevent or delay boundary layer separation. While the method is well reported for gas flows, less knowledge is available for liquid flows. It has been suggested that synthetic jets might be used in hydro turbines in order to stabilise draft tube flow. When accelerating a liquid flow, the energy consumption required will not only depend on the system mass, but also on the added mass due to acceleration of the liquid. The purpose of this thesis has been to study the physics of a synthetic jet where water is the working fluid, drawing special attention to added mass. CFD simulations have been conducted, where the synthetic jet was modelled using a dynamic grid. Results for added mass and frequency response of the system from CFD analysis have been compared to an analytic solution. CFD simulations seem to have captured a number of interesting effects not predicted from the analytic solution; the most prominent being frequency regions of lower added mass than analytically predicted. The apparent resonance regions are believed to occur due to interaction between the excitation frequency of the synthetic jet, and frequencies of dynamics in the system - examples are pressure pulsations from vortex shedding frequencies, or from frequencies originating from the effect of flow across a cavity. Furthermore, it has been demonstrated that operating the system at resonance will reduce the power requirements of the driving force, and that a variable stiffness spring should be part of the actuation system to allow for tuning of the resonance frequency.

Hydroenergi er en norsk produsent av vannturbiner med godt etablert kunnskap innen Kaplan og Francis turbiner. Selskapet har designet flere Pelton skovler med det mål å redusere kavitasjon. Selskapet ønsket å finne et fullstendig Hill-diagram for en av skovlene på en 5 jets Pelton turbin. Andre design skulle bli testet i beste punktet og sammenlignet med nevnte design. I tillegg ønsket de å undersøke optimal design av ringledningen for turbinen. Tester ble utført ved Vannkraftlaboratoriet ved NTNU. Grunnet flere havari i løpet av testperioden ble antall forsøk redusert til kun å finne Hill-diagram som et av designene. Det designet som skulle testes fullt ut ble ødelagt i et av havariene etter at tester med 1 dyse åpen var fullført. De ble erstattet med skovler med samme geometri men med en 2◦ høyere pitch vinkel. Effekten av den økte pitch vinkelen kunne ikke tallfestes. Sammenligning av tester gjort med 1 og 2 dyser indikerer derimot en mulig innvirkning på plasseringen av beste punkter med hensyn til rotasjonshastighet. Et sett av kriterier for optimal design av ringledningen ble funnet og den opprinnelige ringledningen ble redesignet for å møte disse. De redesignede ringledningene gjennomgikk simuleringer i Computational Fluid Dynamics (CFD) basert på tidligere simulering av den opprinnelige ringledningen som er verifisert med eksperimenter. Simuleringene viste en forbedring av falltap ringledningen. I utløpet til dysene ble det observert ugunstig strømning, men dette er antatt å være grunnet feil i modellen. Den hydrauliske virkningsgraden til de testede skovlene ble funnet å være lavere enn på de beste Peltonturbinene på markedet.

Målene med oppgaven var meget sprikende, men alt ledet frem til en endelig design for en kontraroterende aksialturbin som i samarbeid med Lycro A/S skulle bunne ut i en mekanisk design. For å kartlegge de endelige designkriteriene for turbinen skulle det gjennomføres målinger i Førde, hvor falltap og vannføring skulle måles. Disse resultatene skulle brukes for å ferdiggjøre det hydrauliske designverktøyet som var påbegynt under prosjektoppgaven. Videre skulle en akselpakning montert i en tetningsrigg tenkt brukt i turbinen testes, hvor kjølebehovet var av interesse, samt at det skulle foreslås en løsning for kjøling gjennom akselen. I tillegg til dette skulle det ses på en løsning for drenasje av eventuelt lekkasjevann gjennom akselen. Det ble etter hvert klart at deler av oppgaven ville falle bort. Ved en tidligere testkjøring av tetningsriggen gikk denne varm og ble ødelagt, og det var ikke ønske om å gå videre med denne. Samtidig ble feltmålingene utsatt og til slutt droppet på grunn av utsettelser i arbeidet som måtte gjøres i Førde før målingene kunne finne sted. Dette betød at en rekke antakelser måtte gjøres i forhold til designkriteriene for turbinen. Falltapet ble estimert, og for rørgaten som allerede eksisterer i Førde ble det konkludert med at rørdiameteren er alt for liten slik at alt trykket gikk bort i tap, og det ble etter hvert bestemt at et nytt rør må legges. En hydrologisk oversikt ble gjennomført under prosjektoppgaven høsten 2009 for å finne estimert vannføring. Designverktøyet baserer seg på et regneark av Torbjørn Nielsen linket opp mot AutoDesk Inventor. Det ble under prosjektoppgaven lagt vekk på ledeskovldesign, mens det nå er mulig å få et visuelt inntrykk av hele turbinen ut fra designparametere. Mengden kjølevann og behovet for drenering av lekkasjevann ble funnet å være veldig lite, men over tid kan dette bli et stort problem og slite unødig på turbinen. Ved å lage et hull gjennom akselen kan begge problemene løses uten alt for store inngrep i turbindesignet.

Målet med denne oppgaven er å måle pumpevirkningsgrad ved hjelp av termodynamisk metode. Termodynamisk metode måler strømningstapene over en hydraulisk strømningsmaskin. For strømningsmaskiner med løftehøyde over 100 m ligger usikkerheten mellom 0,5-1,0 % i følge ISO 5198. Pumpen målingene er utført på er en neddykket elektrisk sjøvannsløftepumpe, fra Frank Mohn Flatøy AS. Dette er en sentrifugalpumpe med et kort aksielt innløp. Målingene er utført ved Frank Mohn Flatøy AS sin prøvestand. To utfordringer er knyttet til plassering av målepunkt på leveringssiden til denne pumpen. Pumpen er neddykket i en testtank under test, og den drives av en oljefylt elektromotor som er integrert i pumpekonstruksjonen. Som en følge av disse utfordringene er det valgt å benytte flere målepunkter enn det som er vanlig for termodynamisk metode.Pumpens virkningsgrad er også målt med konvensjonell metode hvor tilført elektrisk effekt og avgitt hydraulisk effekt måles. Instrumentering til konvensjonell målemetode er standard instrumentering for testriggen som er brukt. Logging av måledata ble gjennomført i 8 driftspunkt med to logginger i designpunktet.Resultatene fra den termodynamiske målemetoden sammenlignes med resultatene fra konvensjonell målemetode. Måleusikkerheten i den termodynamiske metoden er beregnet.Forventet virkningsgrad er 87 % ut fra tidligere tester med konvensjonell metode på samme pumpe. Den konvensjonelle metoden gir en virkningsgrad på η=87,2 % i beste punkt, mens virkningsgraden fra den termodynamiske metoden er η=77,5 % i beste punkt. Dette er et avvik på 9,7 %. Måleusikkerheten for den termodynamiske metoden er innenfor kravene fra ISO 5198. Ved beste virkningsgradspunkt er måleusikkerheten 0,88 %. Dette tyder på at det er en systematisk feil i den termodynamiske metoden. Den systematiske feilen er antakelig i testoppsettet for temperaturmålepunktet på leveringssiden av pumpen. Entalpien holdes ikke konstant ved trykkreduksjon i målepunktet.

The main objective in this thesis was to investigate a potential increase of power output from the existing turbines at Ruacana Hydropower Station. This was restricted to only consider a runner replacement, and only technical aspects of a rehabilitation project has been evaluated. This involved an investigation of hydraulic condtions, limited to surge oscilliation and water hammer calculations. The results was thereafter used in a calculation model to estimate potential increase in power output. Vital information about the plant and the condition of the waterway was obtained from a study trip to Ruacana. An increase in power output in a turbine can be obtained by increasing the volume flow and/or efficiency. The first criteria can be achieved through expanding the outlet diameter of the runner, and/or altering the flow angles. Therefore has the hydraulic conditions also been analysed with the aim of increasing volume flow. It was found that it can be increased to a maximum nominal volume flow of one unit to 85 m3/s. It was limited by the maximum upsurge in the Surge Headbay with the intake level at highest flood level. The restrictions are altered for the minimum water levels in the intake and maximum levels in the tailrace when operating 1 to 4 units. The advances in Francis runner design over the last deacades makes an increase in efficiency feasible. Finally a calculation method for estimating a potential increase in power output was developed. This was based on traditional turbine theory, and the aim was to find the maximum increase starting from the existing degree of submergence. In addition were the results compared to values obtained from a calculation method developed by Sintef. From a preliminary study of achievements in refurbishment projects was it found that on average has an increase of 18% been obtained. The method developed in this thesis concluded that an increase of 10 - 14% in nominal power output is possble. Assuming a turbine efficiency of 0.93 and only replacing the runner. Sintef's method estimated a maximum value of 15% for the same conditions. It is also found that the outlet diameter at Ruacana can be increased by 60 mm. The additional increase in power output is only estimated to be 1% from the method developed in this thesis. The Sintef method estimates additional 3% increase for the same scenario.

The cross-flow turbine is unique due to the generation of power during two stages. The water flows through the rectangular cross-section nozzle and enters the runner, where the first stage power is generated. The water then flows diametrically through the center of the runner, before it hits the blades on the way out, generating the second stage power. This type of turbine is often used in small hydropower plants located in less-developed countries. The turbine has a simple design, which is economical and easy to manufacture. A cross-flow turbine manufactured by Remote HydroLight in Afghanistan was installed in The Waterpower Laboratory at The Norwegian University of Science and Technology in September 2008. During the fall of 2008, efficiency measurements were performed on the turbine. A maximum efficiency of 78.6% was obtained at 5 meter head. However, although the efficiency is high for a turbine with such a simple design, there is a desire to improve it for better utilization of the resources. An open question is if the flow through the runner behaves like the manufacturers of this turbine type claim. It is therefore of interest to investigate the flow pattern through the runner and the distribution of torque transferred during the two stages. This is the objective of this thesis. Two experiments are performed in this thesis. The objective of the first experiment was to visualize the flow through the runner with use of a high-speed camera. This required an extensive remodeling of the turbine in order to obtain a clear view of the flow. However, the high--speed camera had to be replaced by a single-lens reflex camera and stroboscopes, due to low quality pictures. The second experiment measured the torque transfer to the runner by the use of strain gages. The strain gages could not be calibrated within the time frame of this thesis, but a relative measure of the distribution of torque was obtained. During both experiments the efficiency was measured, but the main objective was to determine the flow pattern and torque transfer through the runner. The results show that the turbine works well for large nozzle openings. The water enters the runner close to the nozzle outlet, leading to a cross flow entering the inside of the runner at a short distance from the nozzle. This gives good conditions for the flow, as the direction of the absolute velocity when entering the second stage corresponds well with the blade inlet angle. At best efficiency point the second stage contributes to 53.7% of the total amount of torque transferred. With decreasing nozzle opening, the cross flow enters the inside of the runner further away from the nozzle. This give a direction of the cross flow which corresponds poorly with the inlet angle of the blades at the second stage, which increases the incidence losses and gives a lower efficiency.

I oppgaven har karakteristikkene for en pumpeturbin blitt simulert. Verktøyet for simuleringene har vært OpenFOAM. Griddene til simuleringene ble laget i Gambit og ble importert til OpenFOAM via fluent gridformat. Geometrien til pumpeturbinen som er benyttet i simuleringene ble stilt til rådighet av Rainpower. Det ble simulert for 28 driftspunkt fordelt på tre forskjellige ledeskovlåpninger. For å begrense simuleringstiden ble kun en av turbinens skovler ble tatt med i beregningsområdet. Resultatene viser god overensstemmelse med måledata for den aktuelle pumpeturbinen, se Figur 1. For å kunne sammenligne resultatene bedre ble det korrigert for tap som er med i måleverdiene, men som ikke omfattes av CFD simuleringene. Konvergens for simuleringer langt fra bestpunkt er problematiske på grunn av områder med tilbakestrømning eller sirkulerende strømning. OpenFOAM versjonene brukt i oppgaven er versjonene 1.5 og 1.5_dev. Versjonen ble byttet underveis fordi de periodiske grensebetingelsene ikke fungerte med versjon 1.5. Figur 1 Volumstrømskarakteristikker

Denne masteroppgaven er skrevet i forbindelse med et pågående doktorgradsarbeid der det blant annet skal gjøres trykkpulsasjonsmålinger på det eksisterende løpehjulet ved Tokke kraftverk. Hensikten med denne trykkpulsasjonsmålingen vil være å teste om det foreslåtte måleoppsettet virker, og å skaffe måledata av eksisterende turbin. Disse dataene vil brukes som sammenligningsgrunnlag når det skal måles på det nye løpehulet som skal installeres på et senere tidspunkt. Resultatene fra arbeidet vil også bli brukt som et grunnlag for senere sammenligning av modell- og fullskalamålinger. Under måling vil det brukes en frittstående datalogger, (CompactRIO) som vil være plassert inne i løpehjulskonusen. Fra dataloggeren vil det gå ledninger ut til miniatyr-trykktransducere som vil være limt fast til løpehjulsskovlene. I forbindelse med dette har måleoppsettet blitt ferdigstilt slik at det kan fungere under testkjøring av turbinen. I hovedsak har dette arbeidet gått ut på å lage praktiske løsninger som gjør det mulig å montere måleoppsettet, samt å forhindre vannlekkasje inn til det elektriske utstyret. I oppgaveteksten var det planlagt gjennomføring av målinger ved Tokke Kraftverk under masterperioden. Dette ble det dessverre ikke gitt mulighet til, grunnet økonomiske årsaker. I stedet ble det, etter samtaler med veileder, valgt å fokusere på å ferdigstille måleoppsettet samt å lage all nødvendig programvare. I tillegg er det kjørt tester for å verifisere at måleoppsettet virker. Siden det viste seg vanskelig å bruke et eksisterende LabView-program, har all nødvendig programvare blitt programmert fra bunnen av. Den programmerte CompactRIO-kontrolleren fungerer på den måten at den ved gitte tidspunkt, som samsvarer med en på forhånd bestemt kjøreplan, vil logge data fra sensorene. Ved de samme tidspunktene vil det bli logget ledeskovlåpning og trykkforhold i sugerørskonusen. Det er også blitt laget verktøy for frekvensanalyse, som vil brukes til frekvensanalyse av de ulike måledataene. I oppgaven er det også gjort tester på programvaren og måleoppsettet. Det største problemet som dukket opp, var begrensninger ved hardwaren på CompactRIOen. Dette medfører at det ved jevne mellomrom må overføres data, og i disse periodene blir det ikke logget data fra sensorene. Rådataene skal frekvensanalyseres, og det er derfor laget et program som tar hensyn til loggepausene ved å legge inn kunstige loggpunkter, basert på snittverdier til hver sensor.