Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Increased use of renewable energy sources and the creation of an internal electricity market have resulted in higher and more volatile power flows in the transmission grid. Due to a sustained climate policy, the share of renewable energy sources in electricity generation will increase over the coming decades. To cope with the expected future power flows, new transmission system investments are inevitable. Over the coming decades, in Europe alone, transmission system investments worth several hundred billion Euro are foreseen.The aim of transmission system planning is to provide an expansion scheme for the transmission network in order to fulfil future grid requirements. Transmission system planning is a complex and multidimensional problem due to the high number of uncertainties and influencing parameters. Therefore, it is difficult to obtain a unique optimal investment plan which satisfies all technical, economical, social and environmental constraints. Many different scenarios regarding generation, demand, market prices, technological and social developments have to be analysed in the planning process in order to assess uncertainties and minimize the investment risks. This dissertation provides the building blocks of a transmission system investment optimization methodology to assess different future scenarios in a fast and effective way. The methodology consists of different modules which can be combined to a single framework or used separately by transmission system planners. The optimization takes the geographic and demographic properties of the area of interest into account which are not included in existing transmission system investment optimization methodologies. It delivers a stepwise transmission system investment plan containing the optimal time point, power rating, transmission route and transmission technology for new investments. Another new feature of the developed methodology is the determination of strong grid nodes based on probabilistic optimal power flow solutions.

621.3 <043> --- Academische collection --- Electrical engineering--Dissertaties --- Theses --- 621.3 <043> Electrical engineering--Dissertaties

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

The need for flexibility within power system operation is growing as more intermittent renewables with limited controllability are integrated. While traditionally this need is met by supply side resources, the demand side also has intrinsic flexibility available which could be tapped, often referred to as demand response. Although policy makers and industry recognize the value of demand response, its use and understanding remains limited. This is especially the case on the residential level.This thesis aims at enhancing the understanding of demand response by addressing three knowledge gaps, ranging from designing dynamic tariff schemes to incentivize demand response, over quantifying the residential load modifications these cause, until determining the final benefits this brings for households and society as a whole.First of all, the demand response incentive following from the current residential tariff designs is limited especially in view of more renewable energy resources. Moreover, these tariffs dont reflect the time-dependency of the underlying cost of electricity. In order to allow demand response and to reflect actual costs to the users, this thesis argues a balance has to be found between tariff principles related to costs and social acceptability on the one hand and its resulting demand response incentive on the other. This balance can be accomplished by proper tariff design. It is shown that the choice of the tariff design not only affects the demand response incentive, but also the resulting benefits.Second, the magnitude to which residential users react to those tariff schemes remains largely unknown. This thesis shows that flexibility obtained from both wet appliances and battery electric vehicles is considerable. Moreover, automation adds to the level and predictability of demand response. Hereby, predictability can be reached by means of price elasticities.Finally, the benefits residential demand response brings to power system operation are not properly identified. This thesis shows that demand response leads to operational benefits as costs of plant operation decrease, while enhancing system reliability. Moreover, demand response proves to be an efficient means to integrate intermittent renewable energy resources. On the investment side, demand response leads to a postponement and reduction of the need for additional generation capacity.

621.3 <043> --- Academische collection --- Electrical engineering--Dissertaties --- Theses --- 621.3 <043> Electrical engineering--Dissertaties

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

The use of large core (1mm) Step Index Plastic Optical Fibers (SI-POF) for last mile and Local Area Network applications is linked to finding an economical solution to the limited bandwidth x length product of the medium. This is mainly caused by the intermodal dispersion which occurs in the fiber. An efficient and economical solution would be the integration of an automatic dispersion correction circuit directly with the optical front-end. This circuit has an equivalent function to adaptive line equalizers, which are already applied widely in existing transmission channels, applied to the specific characteristics of POF.To address the economical reality even further, the integration of this circuit and the optical front-end should be done in a cheap CMOS technology. This introduces additional challenges, both for the implementation of the optical front-end as for the dispersion correction circuit. It are these challenges this PhD seeks to address.

621.3 <043> --- Academische collection --- Electrical engineering--Dissertaties --- Theses --- 621.3 <043> Electrical engineering--Dissertaties

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Recently, themobilemarket has grown tremendously and mobile devices have become ubiquitous and indispensable in our daily life. This introduced a new era for themobile market where the main focus is to reduce the cost and power consumption of battery supplied devices. The power amplifier is a key component in all wireless communication systems. In most of today’s smartphones and other mobile devices, the RF Power Amplifier is predominantly designed in a more exotic technology. To reduce the cost and environmental footprint, it is desirable to completely integrate the RF PA and the entire transceiver into a single system-on-chip (SoC). In addition, the new wireless and mobile communication standards introduce new challenges for fully-integrated power amplifiers. One major challenge is the efficient generation of a Watt-level output power despite the low-breakdown voltage in nanometer scale technologies. As the voltage dropswith the technology scaling, not only the output power and efficiency, but also the stringent requirements on linearity become significantly harder. Certainly, due to the increased data rate, high linearity over instantaneous wide bandwidth is needed in future mobile communication standards. Unlike analog and RF circuitry, digital circuits benefit from technology scaling. Therefore, if a PA is used in switched mode, voltage linearity is not required. A switched-mode PA (SMPA) cannot amplify amplitude modulated (AM) signals, but using techniques such as pulse width modulation (PWM), the amplitude information can be encoded in the time domain and hence efficiently amplified. This thesis manuscript addresses the major challenges of integrating linear and simultaneously highly efficient PAs and PA architectures in standard nanometer technologies in the frequency range from 900 MHz to 5 GHz. In total, four different PAs have been designed: two linear RF PAs, one linear RF PA with integrated power detector and finally a reconfigurable digital RF PA. All RF PAs are designed to cope with several major challenges for fully-integrated RF PA design. The first linear PA is a two-stage power amplifier with a clover-shaped distributed active transformer (DAT) combiner in a standard 90 nm CMOS technology targeting the LTE-communication standard. This test chip achieves the required Watt-level output power for all modern mobile communication standards: an output power of 29.4 dBm is delivered to the antenna with 25.8% power added efficiency (PAE). The RF PA was evaluated with a 10 MHz 16-QAM LTE uplink signal with 7 dB PAPR for LTE-band VIII in the 900-MHz range. Experimental results show that all stringent linearity requirements are satisfied with 15%PAE at an average output power of 25.1 dBm, which is more than the required 23 dBm average output power level. Due to the implementation of a linearity improvement technique, the second RF PA, fabricated in a 180 nm SOI RF PA, is highly linear even over a wide instantaneous bandwidth. This high linearity over a wide bandwidth is required for all future mobile communication standards. The performance of the RF PA is observed for a 64-QAM 20 MHz LTE-advanced uplink signal (PAPR=8.8 dB). This second RF PA delivers a 22.4 dBm with 21.7% PAE, while simultaneously obey all linearity requirements. In addition, the RF PA achieves similar performance at LTE-band I (1.9 GHz) up till LTE-band VII (2.5 GHz) The high linearity of the RF PA allows to transmit a LTE-advanced uplink signal even up to a 60 MHz instantaneous bandwidth with a 10.3 dB PAPR efficiently. Since the power amplifier is the most power hungry building block in a transmitter, the third linear RF PA is integrated together with an on-chip power detector in a 40 nm standard CMOS technology. The on-chip power detector detects both the RF output current and the RF output voltage to monitor the output power of the RF PA and hence is able to measure the output power even for non-fixed 50 Ohm load. The power detector introduces a sense winding to measure the RF output current. The RF PA is a single stage RF PA using transformer-based input and output matching to support the 5 GHz WLAN IEEE802.11ac communication standard. With continuous wave signals, the RF PA achieves 38.8%PAE and a drain efficiency of 49.9%at a 24 dBm output power. With a 40 MHz 64-QAM WLAN signal at 5 GHz, the RF PA transmits 16.9 dBm while operating with 17.3% efficiently. Finally, a burst-mode RF transmitter was presented to achieve efficiency enhancement. A fully-integrated Burst Mode transmitter has been realized using aforementioned fully integratedWatt-level RF PA togetherwith a baseband PWMmodulator processed in 65 nm standard CMOS, designed by P. Nuyts. During Burst-Mode operation, the RF PA operates always in saturation. And at 930 MHz, it has a gain of 31 dB at low duty cycles and drops to 29 dB at peak output power. It implies that the variation over the whole duty cycle range is limited to 2 dB. This proves the inherent linear behavior of the burst mode amplifier. At 100% duty cycle, the power amplifier delivers a peak output power of 28.8 dBm with 23.7% PAE. While applying a 20 MHz WLAN IEEE802.11b/g to the PA, the RF PA achieves 11.7% PAE at 23.1 dBm average output power, where the EVM limit is reached. The transmitter clearly satisfies the industry-based LTE EVM specification of 5.6% for output powers ranging from about 14 to 23 dBm. This proved the potential for a single chip design. Consequently, a fourth single chip fully-integrated reconfigurable multilevel PWM transmitter was designed in a 40 nm standard CMOS. To fully-digital multi-level RF PA is implemented in a 40 nm standard CMOS process to generate a multi-level PWM signal and at the same time achieve the required watt-level output power for modern mobile systems. This RF PA architecture uses transformer-based series power combiner to sum the output power of several power amplifiers. Due to the multi-mode operation, simulations show still 3.5 times more efficiency at 15 dB back-off than a conventional Class B amplifier. Simultaneously, this multi-mode RF PA extends the dynamic range by 15 dB and hence doesn’t constrain this digital transmitter architecture to modulated signals with only low PAPR. Due to technical issues, experiments could never confirm the multi-level operation nor the 3.5 times more efficiency than a conventional Class B at 15 dB back-off.

681.3 <043> --- Academische collection --- Computer science--Dissertaties --- Theses

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Multidimensional data, or tensors, arise natura lly in data analysis applications. Hitchcock&##39;s ¨rank decomposition is one of several tensor decom positions that can reveal a regular structure unde rlying such data, and it has found application in¨ algebraic statistics, chemometrics, machine learni ng and signal processing, among others. This rank¨ decomposition can be regarded as a generalization¨ of the singular value decomposition ofsecond-order ¨tensors, i.e., matrices. Contrary to the case of¨ matrices,Hitchcock decompositions of tensors of or der at least three are not yetthoroughly comprehen ded at present. Therefore, in the first part of th is thesis, several theoretical properties of the r ank decomposition that are of practical significan ce are investigated. The dimension of the setof Hi tchcock decompositions of a fixed rank is studied¨ with a randomized algorithm that provides probabil istic statements about this dimension.Since the di mension is usually as expected, a more refined inq uiry intothe uniqueness of Hitchcock decomposition s is undertaken. An efficient algorithm is propose d for proving that a generic tensor of low rank ad mits a unique Hitchcock decomposition; it is addit ionally extended to certify uniqueness of a given¨ decomposition. Thereafter, as a consequence ofthe¨ foregoing results, it is shown that a Hitchcock de composition of a generic tensor of low rank cannot ¨be truncated optimally, in contrast tothe case of ¨matrices. It is demonstrated, moreover, that a ra nk decomposition, unfortunately, cannot, in genera l, be computed through a simple greedy algorithm t hat computes successive rank-1 approximations of t he tensor. The second part of the thesis develops¨ computational kernels associated with rank decompo sitions. The higher-order singular value decomposi tion is often employed as a means for reducing the ¨computational complexity of algorithms for comput ing Hitchcock&##39;s decomposition. An alternative¨ algorithm for computing the former, which is more¨ efficient than theclassic approach, is proposed. A ¨strategy for efficiently truncating itis then der ived from this alternative algorithm. Finally, an¨implementation of the computation of the gradient¨ of the objective function associated with the rank ¨decomposition is developed; it attains a throughp ut that is close to the theoretical limit of the c omputer system while employing only a constant amo unt of memory, hereby improving upon the currentst ate-of-the-art algorithms.

681.3 <043> --- Academische collection --- Computer science--Dissertaties --- Theses