The trend in the electronic and electrical equipment industry towards denser and more powerful product requires a higher level of performance from cooling devices. In this context, passive cooling techniques such as latent heat storage systems have attracted considerable attention in recent years. Phase change materials (PCMs) have turned out to be extremely advantageous in this regard as they absorb high amount of latent heat without much rise of temperature. But unfortunately, nearly all phase change materials (PCMs) with high latent heat storage capacity have unacceptably low thermal conductivity, which makes heating and cooling processes slow during melting and solidification of PCMs. Augmentation of heat transfer in a PCM is achieved by inserting a high thermal conductivity material, known as thermal conductivity enhancer (TCE), into the PCM. The conglomeration of PCM and TCE is known as a thermal storage unit (TSU). In this thesis, detailed and systematic analyses are presented on the thermal performance of TSUs subjected to two types of thermal loading- (a) constant power loading in which a constant power level is supplied to the chip (heater) for a limited duration of time, and (b) cyclic loading. Eicosane is used as the PCM, while aluminium pin or plate fins are used as TCEs. First, a 1-D analytical model is developed to obtain a closed-form temperature distribution for a simple PCM domain (without TCE) heated uniformly from the bottom. The entire heating process is divided into three stages, viz. (a) sensible heating period before melting, during which heat is stored in the solid PCM in the form of specific heat, (b) melting period, during which a melt front progresses from the bottom to the top layer of the PCM and heat is stored in latent as well as in sensible forms, and (c) post melting period, during which energy is stored again in the form of sensible heat. For each stage, conduction energy equation is solved with a set of initial and boundary conditions. Subsequently, a resistance capacitance model of phase change process is developed for further analysis. For transient performance under constant thermal loading, experimental investigations are carried out for TSUs with different percentages of TCE. A numerical model is developed to interpret the experimental results. The thermal performance of a TSU is found to depend on a number of geometrical parameters and boundary conditions. Hence, a systematic approach is desirable for finding the best TSU design for which the chip can be operated for a longer period of time before it reaches a critical temperature (defined as the temperature above which the chip starts malfunctioning). As a first step of the approach, it is required to identify the parameters which can affect the transient process. It is found that the convective heat transfer coefficient, ‘h’ and the exposed area for heat transfer have little effect on the chip temperature during the constant power operation. A randomized search technique, Genetic Algorithm (GA), is coupled with the CFD code to find an optimum combination of geometrical parameters of TSUs based on the design criteria. First, the optimization is carried out without considering melt convection within the PCM. It is found that the optimum half-fin width remains fixed for a given heat flux and temperature difference. Assuming a quasi steady process, the results of optimization are then explained by constructing and analyzing a resistance network model. The resistance network model is then extended to include the effect of melt convection, and it is shown that the optimum pitch changes with the strength of convection. Accordingly, numerical analysis is carried out by considering the effect of melt convection, and a correlation for optimum pitch is developed. Having established the role of melt convection on the thermal performance of TSUs, rigorous computational and experimental studies are performed in order to develop correlations among different non-dimensional numbers, such as Nusselt number, Rayleigh number, Stefan number and Fourier number, based on a characteristic length scale for convection. The enclosures are classified into three types, depending on the aspect ratio of cavity, viz. shallow, rectangular and tall enclosures. For a shallow enclosure, the characteristic length is the height of cavity whereas for a tall enclosure, the characteristic length is the fin pitch. In case of rectangular enclosure, both pitch and height are the important characteristic lengths. For cyclic operation, it is required that the fraction of the PCM melting during the heating cycle should completely solidify back during the cooling period, in order that that TSU can be operated for an unlimited number of cycles. If solidification is not complete during the cooling period, the TSU temperature will tend to rise with every cycle, thus making it un-operational after some cycles. It is found that the solidification process during the cooling period depends strongly on the heat transfer coefficient and the cooling surface area. However, heat transfer coefficient does not play any significant role during the heating period; hence a TSU optimized for transient operation may not be ideal for cyclic loading. Accordingly, studies are carried out to find the parameters which could influence the behaviour of PCM under cyclic loading. A number of parameters are identified in the process, viz. cycle period and heat transfer coefficient. It is found that the required heat transfer coefficient for infinite cyclic operation is very high and unrealistic with air cooling from the surface of the TSU. Otherwise, the required cooling period for complete re-solidification will be very high, which may not be suitable for most applications. In an effort to bring down the cooling period to a duration that is comparable to the heating period, a new design is proposed where both ‘h’ and area exposed to heat transfer can be controlled. In this new design, the gaps between the fins in a plate-fin TSU are alternately filled with PCM, such that only one side of a fin is in contact with PCM and the other side is exposed to the coolant (air). In this arrangement, the same heat flow path through the fin which is used for heating the PCM (during the heating stage) can also be used for cooling and solidifying the PCM during the cooling part of the cycle. Natural or forced air cooling through the passages can be introduced to provide a wide range of heat transfer coefficient which can satisfy the cooling requirements. With this arrangement, the enhanced area provided for cooling keeps the ‘h’ requirement within a realistic limit. This cooling method developed is categorized as a combination of active and passive cooling techniques. Analytical and numerical investigations are carried out to evaluate the thermal performance of this modified PCM-based heat sink in comparison to the ones with conventional designs. It is found that, the performance of new PCM-based heat sink is superior to that of the conventional one. Experiments are performed on both the conventional and the new PCM-based heat sinks to validate the new findings.