Dissertations / Theses on the topic 'Energy efficiency not elsewhere classified'

Molecular separations are essential in the production of many chemicals and purified products. Of all the available separation technologies, distillation, which is a thermally driven process, has been and continues to be one of the most utilized separation methods in chemical and petrochemical plants. Although distillation and other commercial technologies fulfilled most of the current separation needs, the energy-intensive nature of many molecular separations and the growing concern of reducing CO₂ emissions has led to intense research to seek for more energy-efficient separation processes.

Among the emerging separation technologies alternative to distillation, there is special attention on non-thermally driven methods, such as membranes. The growing interest in non-thermal methods, and particularly in the use of membranes, has been influenced significantly from the widespread perception that they have a potential to be markedly less energy-intensive than thermal methods such as distillation. Even though many publications claim that membranes are more energy-efficient than distillation, except for water desalination, the relative energy intensity between these processes in the separation of chemical mixtures has not been deeply studied in the literature. One of the objectives of this work focuses on introducing a framework for comparative analysis of the energy intensity of membranes and distillation.

A complication generally encountered when comparing the energy consumption of membranes against an alternative process is that often the purity and recovery that can be achieved through a single membrane stage is limited. While using a multi-stage membrane process is a plausible solution to achieve both high purity and recovery, even for a simple binary separation, finding the most suitable multistage membrane process is a difficult task. This is because, for a given separation, there exists multiple cascades that fulfill the separation requirements but consume different amounts of energy. Moreover, the energy requirement of each cascade depends on the operating conditions. The first part of this work is dedicated to the development of a Mixed Integer Non-linear Program (MINLP) which allows for a given gaseous or liquid binary separation, finding the most energy-efficient membrane cascade. The permeator model, which is derived from a combination of the cross-flow model and the solution diffusion theory, and is originally expressed as a differential-algebraic equation (DAE) system, was integrated analytically before being incorporated in the optimization framework. This is in contrast to the common practice in the literature, where the DAE system is solved using various discretization techniques. Since many of the constraints have a non-convex nature, local solvers could get trapped in higher energy suboptimal solutions. While an option to overcome this limitation is to use a global solver such as BARON, it fails to solve the MINLP to the desired optimality in a reasonable amount of time for most of the cases. For this reason, we derive additional cuts to the problem by exploiting the mathematical properties of the governing equations and from physical insights. Through numerical examples, we demonstrate that the additional cuts aid BARON in expediting the convergence of branch-and-bound and solve the MINLP within 5%-optimality in all the cases tested in this work.

The proposed optimization model allows identifying membrane cascades with enhanced energy efficiency that could be potentially used for existing or new separations. In addition, it allows to compare the optimum energy consumption of a multistage membrane process against alternative separations methods and aid in the decision of whether or not to use a membrane system. Nevertheless, it should be noted that when a membrane process or any other non-thermal separation process is compared with a thermal process such as distillation, an additional complication often arises because these processes usually use different types of energies. Non-thermal processes, such as membranes, consume electrical energy as work, whereas thermal processes, such as distillations, usually consume heat, which is available in a wide range of temperatures. Furthermore, the amount of fuel consumed by a separation process strongly depends on how its supplied energy is produced, and how it is energy integrated with the rest of the plant. Unfortunately, common approaches employed to compare the energy required by thermal and non-thermal methods often lead to incorrect conclusions and have driven to the flawed perception that thermal methods are inherently more energy-intensive than non-thermal counterparts. In the second part of this work, we develop a consistent framework that enables a proper comparison of the energy consumption between processes that are driven by thermal and non-thermal energy (electrical energy). Using this framework, we refute the general perception that thermal separation processes are necessarily the most energy-intensive and conclusively show that in several industrially important separations, distillation processes consume remarkably lower fuel than non-thermal membrane alternatives, which have often been touted as more energy efficient.

In order to gain more understanding of the conditions where membranes or distillation are more energy-efficient, we carried out a comprehensive analysis of the energy consumed by these two processes under different operating conditions. The introduced energy comparison analysis was applied to two important separation examples; the separation of p-xylene/o-xylene, and propylene/propane. Our results showed that distillation is more energy favored than membranes when the target purity and recovery of the most volatile (resp. most permeable) component in the distillate (resp. permeate) are high, and particularly when the feed is not too concentrated in the most volatile (resp. most permeable) component. On the other hand, when both the recovery and purity of the most volatile (resp. most permeable) component are required at moderate levels, and particularly when the feed is highly enriched in the most volatile (resp. most permeable) component, membranes show potential to save energy as compared to distillation.

Modern computing platforms, from servers to mobile devices, demand ever-increasing amounts of memory to keep up with the growing amounts of data they process, and to bridge the widening processor-memory gap. A large and growing fraction of chip area and energy is expended in memories, which face challenges with technology scaling due to increased leakage, process variations, and unreliability. On the other hand, data intensive workloads such as machine learning and data analytics pose increasing demands on memory systems. Consequently, improving the energy-efficiency and performance of memory systems is an important challenge for computing system designers.

Spintronic memories, which offer several desirable characteristics - near-zero leakage, high density, non-volatility and high endurance - are of great interest for designing future memory systems. However, these memories are not drop-in replacements for current memory technologies, viz. Static Random Access Memory (SRAM) and Dynamic Random Access Memory (DRAM). They pose unique challenges such as variable access times, and require higher write latency and write energy. This dissertation explores new approaches to improving the energy efficiency of spintronic memory systems.

The dissertation first explores the design of approximate memories, in which the need to store and access data precisely is foregone in return for improvements in energy efficiency. This is of particular interest, since many emerging workloads exhibit an inherent ability to tolerate approximations to their underlying computations and data while still producing outputs of acceptable quality. The dissertation proposes that approximate spintronic memories can be realized either by reducing the amount of data that is written to/read from them, or by reducing the energy consumed per access. To reduce memory traffic, the dissertation proposes approximate memory compression, wherein a quality-aware memory controller transparently compresses/decompresses data written to or read from memory. For broader applicability, the quality-aware memory controller can be programmed to specify memory regions that can tolerate approximations, and conforms to a specified error constraint for each such region. To reduce the per-access energy, various mechanisms are identified at the circuit and architecture levels that yield substantial energy benefits at the cost of small probabilities of read, write or retention failures. Based on these mechanisms, a quality-configurable Spin Transfer Torque Magnetic RAM (STT-MRAM) array is designed in which read/write operations can be performed at varying levels of accuracy and energy at runtime, depending on the needs of applications. To illustrate the utility of the proposed quality-configurable memory array, it is evaluated as an L2 cache in the context of a general-purpose processor, and as a scratchpad memory for a domain-specific vector processor.

The dissertation also explores the design of caches with Domain Wall Memory (DWM), a more advanced spintronic memory technology that offers unparalleled density arising from a unique tape-like structure. However, this structure also leads to serialized access to the bits in each bit-cell, resulting in increased access latency, thereby degrading overall performance. To mitigate the performance overheads, the dissertation proposes a reconfigurable DWM-based cache architecture that modulates the active bits per tape with minimal overheads depending on the application's memory access characteristics. The proposed cache is evaluated in a general purpose processor and improvements in performance are demonstrated over both CMOS and previously proposed spintronic caches.

In summary, the dissertation suggests directions to improve the energy efficiency of spintronic memories and re-affirms their potential for the design of future memory systems.

Cyber physical systems are deployed in a wide range of applications from sensor nodes in a factory setting to drones in defense applications. This distributed setting of nodes or processes often needs to reach agreement on a set of values. Byzantine Agreement protocols address this issue of reaching an agreement in an environment where a malicious entity can take control over a set of nodes and deviates the system from its normal operation. However these protocols do not consider the energy consumption of the nodes. We explore Byzantine Agreement protocols from an energy efficient perspective providing both energy resilience where the actions of the Byzantine nodes can not adversely effect the energy consumption of non-malicious nodes as well as fairness in energy consumption of nodes over multiple rounds of agreement.

Spiking Neural Networks (SNNs), widely known as the third generation of artificial neural networks, offer a promising solution to approaching the brains' processing capability for cognitive tasks. With more biologically realistic perspective on input processing, SNN performs neural computations using spikes in an event-driven manner. The asynchronous spike-based computing capability can be exploited to achieve improved energy efficiency in neuromorphic hardware. Furthermore, SNN, on account of spike-based processing, can be trained in an unsupervised manner using Spike Timing Dependent Plasticity (STDP). STDP-based learning rules modulate the strength of a multi-bit synapse based on the correlation between the spike times of the input and output neurons. In order to achieve plasticity with compressed synaptic memory, stochastic binary synapse is proposed where spike timing information is embedded in the synaptic switching probability. A bio-plausible probabilistic-STDP learning rule consistent with Hebbian learning theory is proposed to train a network of binary as well as quaternary synapses. In addition, hybrid probabilistic-STDP learning rule incorporating Hebbian and anti-Hebbian mechanisms is proposed to enhance the learnt representations of the stochastic SNN. The efficacy of the presented learning rules are demonstrated for feed-forward fully-connected and residual convolutional SNNs on the MNIST and the CIFAR-10 datasets.

STDP-based learning is limited to shallow SNNs (<5 layers) yielding lower than acceptable accuracy on complex datasets. This thesis proposes block-wise complexity-aware training algorithm, referred to as BlocTrain, for incrementally training deep SNNs with reduced memory requirements using spike-based backpropagation through time. The deep network is divided into blocks, where each block consists of few convolutional layers followed by an auxiliary classifier. The blocks are trained sequentially using local errors from the respective auxiliary classifiers. Also, the deeper blocks are trained only on the hard classes determined using the class-wise accuracy obtained from the classifier of previously trained blocks. Thus, BlocTrain improves the training time and computational efficiency with increasing block depth. In addition, higher computational efficiency is obtained during inference by exiting early for easy class instances and activating the deeper blocks only for hard class instances. The ability of BlocTrain to provide improved accuracy as well as higher training and inference efficiency compared to end-to-end approaches is demonstrated for deep SNNs (up to 11 layers) on the CIFAR-10 and the CIFAR-100 datasets.

Feed-forward SNNs are typically used for static image recognition while recurrent Liquid State Machines (LSMs) have been shown to encode time-varying speech data. Liquid-SNN, consisting of input neurons sparsely connected by plastic synapses to randomly interlinked reservoir of spiking neurons (or liquid), is proposed for unsupervised speech and image recognition. The strength of the synapses interconnecting the input and liquid are trained using STDP, which makes it possible to infer the class of a test pattern without a readout layer typical in standard LSMs. The Liquid-SNN suffers from scalability challenges due to the need to primarily increase the number of neurons to enhance the accuracy. SpiLinC, composed of an ensemble of multiple liquids, where each liquid is trained on a unique input segment, is proposed as a scalable model to achieve improved accuracy. SpiLinC recognizes a test pattern by combining the spiking activity of the individual liquids, each of which identifies unique input features. As a result, SpiLinC offers comparable accuracy to Liquid-SNN with added synaptic sparsity and faster training convergence, which is validated on the digit subset of TI46 speech corpus and the MNIST dataset.

With the Moore’s Law era coming to an end, much attention has been given to novel nanoelectronic devices as a key driving force behind technological innovation. Utilizing the inherent device physics of nanoelectronic components, for sensory and computational tasks have proven to be useful in reducing the area and energy requirements of the underlying hardware fabrics. In this work we demonstrate how the intrinsic noise present in nano magnetic devices can pave the pathway for energy efficient neuromorphic hardware. Furthermore, we illustrate how the unique magnetic properties of such devices can be leveraged for accurate estimation of environmental magnetic fields. We focus on spintronic technologies in particular, due to the low current and energy requirements in contrast to traditional CMOS technologies.

Image segmentation is a crucial pre-processing stage used in many object identification tasks that involves simplifying the representation of an image so it can be conveniently analyzed in the later stages of a problem. This is achieved through partitioning a complicated image into specific groups based on color, intensity or texture of the pixels of that image. Locally Excitatory Globally Inhibitory Oscillator Network or LEGION is one such segmentation algorithm, where synchronization and desynchronization between coupled oscillators are used for segmenting an image. In this work we present an energy efficient and scalable hardware implementation of LEGION using stochastic Magnetic Tunnel Junctions that leverage the fast parallel

nature of the algorithm. We demonstrate that the proposed hardware is capable of segmenting binary and gray-scale images with multiple objects more efficiently than
existing hardware implementations.

It is understood that the underlying device physics of spin devices can be used for emulating the functionality of a spiking neuron. Stochastic spiking neural networks based on nanoelectronic spin devices can be a possible pathway of achieving brain-like compact and energy-efficient cognitive intelligence. Current computational models attempt to exploit the intrinsic device stochasticity of nanoelectronic synaptic or neural components to perform learning and inference. However, there has been limited analysis on the scaling effect of stochastic spin devices and its impact on the operation of such stochastic networks at the system level. Our work attempts to explore the design space and analyze the performance of nanomagnet based stochastic neuromorphic computing architectures, for magnets with different barrier heights. We illustrate how the underlying network architecture must be modified to account for the random telegraphic switching behavior displayed by magnets as they are scaled into the superparamagnetic regime.

Next we investigate how the magnetic properties of spin devices can be utilized for real world sensory applications. Magnetic Tunnel Junctions can efficiently translate variations in external magnetic fields into variations in electrical resistance. We couple this property of Magnetic Tunnel Junctions with Amperes law to design a non-invasive sensor to measure the current flowing through a wire. We demonstrate how undesirable effects of thermal noise and process variations can be suppressed through novel analog and digital signal conditioning techniques to obtain reliable and accurate current measurements. Our results substantiate that the proposed noninvasive current sensor surpass other state-of-the-art technologies in terms of noise and accuracy.

Research on net-zero energy buildings (NZEBs) has been largely centered around improving building energy performance, while little attention has been given to indoor air quality. A critically important class of indoor air pollutants are nanoaerosols – airborne particulate matter smaller than 100 nm in size. Nanoaerosols penetrate deep into the human respiratory system and are associated with deleterious toxicological and human health outcomes. An important step towards improving indoor air quality in NZEBs is understanding how occupants, their activities, and building systems affect the emissions and fate of nanoaerosols. New developments in smart energy monitoring systems and smart thermostats offer a unique opportunity to track occupant activity patterns and the operational status of residential HVAC systems. In this study, we conducted a one-month field campaign in an occupied residential NZEB, the Purdue ReNEWW House, to explore how energy use profiles and smart thermostat data can be used to characterize indoor nanoaerosol dynamics. A Scanning Mobility Particle Sizer and Optical Particle Sizer were used to measure indoor aerosol concentrations and size distributions from 10 to 10,000 nm. AC current sensors were used to monitor electricity consumption of kitchen appliances (cooktop, oven, toaster, microwave, kitchen hood), the air handling unit (AHU), and the energy recovery ventilator (ERV). Two Ecobee smart thermostats informed the fractional amount of supply airflow directed to the basement and main floor. The nanoaerosol concentrations and energy use profiles were integrated with an aerosol physics-based material balance model to quantify nanoaerosol source and loss processes. Cooking activities were found to dominate the emissions of indoor nanoaerosols, often elevating indoor nanoaerosol concentrations beyond 10⁴ cm^-3. The emission rates for different cooking appliances varied from 10¹¹ h^-1 to 10¹⁴ h^-1. Loss rates were found to be significantly different between AHU/ERV off and on conditions, with median loss rates of 1.43 h^-1 to 3.68 h^-1, respectively. Probability density functions of the source and loss rates for different scenarios will be used in Monte Carlo simulations to predict indoor nanoaerosol concentrations in NZEBs using only energy consumption and smart thermostat data.

Breast cancer is the predominant cancer diagnosed among women, and the second most deadly cancer. The vast majority of cancer-related deaths is caused by the metastatic spread of cancer from the primary tumor to a distant site in the body. Therefore, new strategies which minimize breast cancer metastasis are imperative to improve patient survival. Cancer cells which acquire anchorage independence, or the ability to survive without extracellular matrix attachment, and metabolic flexibility have increased potential to metastasize. In the present studies, the ability to survive detachment and subsequent metabolic changes were determined in human Harvey-ras transformed MCF10A-ras breast cancer cells. Detachment resulted in reduced viability in a time-dependent manner with the lowest cell viability observed at forty hours. In addition, decreased cell viability was observed in both glutamine and glucose depleted detached conditions, suggesting a dependence on both nutrients for detached survival. Compared to attached cells, detached cells had reduced total pool sizes of pyruvate, lactate, α-ketoglutarate, fumarate, malate, alanine, serine, and glutamate, suggesting the metabolic stress which occurs under detached conditions. However, intracellular citrate and aspartate pools were unchanged, demonstrating a preference to maintain these pools in detached conditions. Compared to attached cells, detached cells had suppressed glutamine metabolism, as determined by decreased glutamine flux into the TCA cycle and reduced mRNA abundance of glutamine metabolizing enzymes. Further, detached glucose anaplerosis through pyruvate dehydrogenase activity was decreased, while pyruvate carboxylase (PC) expression and activity were increased. A switch in metabolism was observed away from glutamine anaplerosis to a preferential utilization of PC activity to replenish the TCA cycle, determined by reduced PC mRNA abundance in detached cells treated with a cell-permeable analog of α-ketoglutarate, the downstream metabolite of glutamine which enters the TCA cycle. These results suggest that detached cells elevate PC to increase flux of carbons into the TCA cycle when glutamine metabolism is reduced.

Vitamin D is recognized for its role in preventing breast cancer progression, and recent studies suggest that regulation of energy metabolism may contribute to its anticancer effects. Vitamin D primarily acts on target tissue through its most active metabolite, 1α,25-dihydroxyvitamin D (1,25(OH)₂D). The present work investigated 1,25(OH)₂D’s effects on viability of detached cells through regulation of energy metabolism. Treatment of MCF10A-ras cells with 1,25(OH)₂D resulted in decreased viability of detached cells. While 1,25(OH)₂D treatment did not affect many of the glucose metabolism outcomes measured, including intracellular pyruvate and lactate pool sizes, glucose flux to pyruvate and lactate, and mRNA abundance of enzymes involved in glucose metabolism, 1,25(OH)₂D treatment reduced detached PC expression and glucose flux through PC. A reduction in glutamine metabolism was observed with 1,25(OH)₂D treatment, although no 1,25(OH)₂D target genes were identified. Further, PC depletion by shRNA decreased cell viability in detached conditions with no additional effect with 1,25(OH)₂D treatment. Moreover, PC overexpression resulted in increased detached cell viability and inhibited 1,25(OH)₂D’s negative effects on viability. These results suggest that 1,25(OH)₂D reduces detached cell viability through regulation of PC. Collectively this work identifies a key metabolic adaptation where detached cells increase PC expression and activity to compensate for reduced glutamine metabolism and that 1,25(OH)₂D may be utilized to reverse this effect and decrease detached cell viability. These results contribute to an increased understanding of metastatic processes and the regulation of these processes by vitamin D, which may be effective in preventing metastasis and improve breast cancer patient survival.