Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “heuristic”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 613 records · Page 34

Reinforcement learning building control approach harnessing imitation learning

Reinforcement learning (RL) has shown significant success in sequential decision making in fields like autonomous vehicles, robotics, marketing and gaming industries. This success has attracted the attention to the RL control approach for building energy systems which are becoming complicated due to the need to optimize for multiple, potentially conflicting, goals like occupant comfort, energy use and grid interactivity. However, for real world applications, RL has several drawbacks like requiring large training data and time, and unstable control behavior during the early exploration process making it infeasible for an application directly to building control tasks. To address these issues, an imitation learning approach is utilized herein where the RL agents starts with a policy transferred from accepted rule based policies and heuristic policies. This approach is successful in reducing the training time, preventing the unstable early exploration behavior and improving upon an accepted rule-based policy - all of these make RL a more practical control approach for real world applications in the domain of building controls.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Effective detection of CO 2 leakage: a comparison of groundwater sampling and pressure monitoring

Shallow aquifer monitoring is likely to be a required aspect to any geologic CO 2 sequestration operation. Collecting groundwater samples and analyzing for geochemical parameters such as pH, alkalinity, total dissolved carbon, and trace metals has been suggested by a number of authors as a possible strategy to detect CO 2 leakage. The effectiveness of this approach, however, will depend on the hydrodynamics of the leak-induced CO 2 plume and the spatial distribution of the monitoring wells relative to the origin of the leak. To our knowledge, the expected effectiveness of groundwater sampling to detect CO 2 leakage has not yet been quantitatively assessed. In this study we query hundreds of simulations developed for the National Risk Assessment Project (US DOE) to estimate risks to drinking water resources associated with CO 2 leaks. The ensemble of simulations represent transient, 3-D multi-phase reactive transport of CO 2 and brine leaked from a sequestration reservoir, via a leaky wellbore, into an unconfined aquifer. Key characteristics of the aquifer, including thickness, mean permeability, background hydraulic gradient, and geostatistical measures of aquifer heterogeneity, were all considered uncertain parameters. Complex temporally-varying CO 2 and brine leak rate scenarios were simulated using a heuristic scheme with ten uncertain parameters. The simulations collectively predict the spatial and temporal evolution of CO 2 and brine plumes over 200 years in a shallow aquifer under a wide range of leakage scenarios and aquifer characteristics. Using spatial data from an existing network of shallow drinking water wells in the Edwards Aquifer, TX, as one illustrative example, we calculated the likelihood of leakage detection by groundwater sampling. In this monitoring example, there are 128 wells available for sampling, with a density of about 2.6 wells per square kilometer. If the location of the leak is unknown a priori, a reasonable assumption in many cases, we found that the leak would be detected in at least one of the monitoring wells in less than 10% of the scenarios considered. This is because plume sizes are relatively small, and so the probability of detection decreases rapidly with distance from the leakage point. For example, 400m away from the leakage point there is less than 20% chance of detection. We then compared the effectiveness of groundwater quality sampling to shallow aquifer and/or reservoir pressure monitoring. For the Edwards Aquifer example, pressure monitoring in the same monitoring well network was found to be even less effective that groundwater quality monitoring. This is presumably due to the unconfined conditions and relatively high permeability, so pressure perturbations quickly dissipate. Although specific results may differ from site to site, this type of analysis should be useful to site operators and regulators when selecting leak detection strategies. Given the spatial characteristics of a proposed monitoring well network, probabilities of leakage detection can be rapidly calculated using this methodology. Although conditions such as these may not be favorable for leakage detection in shallow aquifers, leakage detection could be much more successful in the injection reservoir. We demonstrate proof-of-concept for this hypothesis, presenting a simulation where there is measurable pressure change at the injection well due to overpressurization, fault rupture, and consequent leakage up the fault into intermediate and shallow aquifers. The size of the detectible pressure change footprint is much larger in the reservoir than in either of the overlying aquifers. Further exploration of the range of conditions for which this technique would be successful is the topic of current study.

54 ENVIRONMENTAL SCIENCES↗

Governance in Strategic Environmental Assessment: Lessons from the Portuguese practice

The analysis of governance in Strategic Environmental Assessment (SEA) can help understand why, whether and how strategic decision-making happens. Understanding the governance context is strategic to improve the role and capacity of SEA to stimulate, and legitimate decisions that integrate environmental issues and are sustainability driven. The objective of this paper is to discuss why governance is important in SEA. In the SEA literature governance is mostly addressed in silos (i.e. public participation or decisions transparency or accountability) rather than in an integrated way. In addition few authors adopt a strategic view to address the governance context within which SEA is used. In this paper we address the heuristics of governance in SEA based on theoretical and empirical evidence, suggesting how SEA may incorporate the governance dimension. First a review of the SEA literature in relation to governance sets the context to the analysis on how governance is approached in practice, based on 60 Portuguese SEA cases. This is followed by the presentation of an empirical SEA case conducted in Portugal to illustrate what, in our understanding, can be an example of good practice in considering governance in SEA. Final discussion reflects on the role of governance in SEA in promoting engagement, enabling collaborative action, learning processes and dialogues, concluding on the relevance of governance in creating development contexts that can deal with change.

54 ENVIRONMENTAL SCIENCES↗

Mixed-integer programming models for optimal constellation scheduling given cloud cover uncertainty

We introduce the problem of scheduling observations on a constellation of remote sensors, to maximize the aggregate quality of the collections obtained. While automated tools exist to schedule remote sensors, they are often based on heuristic scheduling techniques, which typically fail to provide bounds on the quality of the resultant schedules. To address this issue, we first introduce a novel deterministic mixed-integer programming (MIP) model for scheduling a constellation of one to n satellites, which relies on extensive pre-computations associated with orbital propagators and sensor collection simulators to mitigate model size and complexity. Our MIP model captures realistic and complex constellation-target geometries, with solutions providing optimality guarantees. We then extend our base deterministic MIP model to obtain two-stage and three-stage stochastic MIP models that proactively schedule to maximize expected collection quality across a set of scenarios representing cloud cover uncertainty. Our experimental conclusions on instances of one and two satellites demonstrate that our stochastic MIP models yield significantly improved collection quality relative to our base deterministic MIP model. We further demonstrate that commercial off-the-shelf MIP solvers can produce provably optimal or near-optimal schedules from these models in time frames suitable for sensor operations.

47 OTHER INSTRUMENTATION↗

Aggregation and data driven identification of building thermal dynamic model and unmeasured disturbance

An aggregate model is a single-zone equivalent of a multi-zone building, and is useful for many purposes, including model based control of large heating, ventilation and air conditioning (HVAC) equipment. This paper deals with the problem of simultaneously identifying an aggregate thermal dynamic model and unknown disturbances from input–output data of multi-zone buildings. The unknown disturbance is a key challenge since it is not measurable but non-negligible. In this paper, we first present a principled method to aggregate a multi-zone building model into a single zone model, and show the aggregation is not as trivial as it has been assumed in the prior art. We then provide a method to identify the parameters of the model and the unknown disturbance for this aggregate (single-zone) model. Finally, we test our proposed identification algorithm to data collected from a multi-zone building testbed in Oak Ridge National Laboratory. A key insight provided by the aggregation method allows us to recognize under what conditions the estimation of the disturbance signal will be necessarily poor and uncertain, even in the case of a specially designed test in which the disturbances affecting each zone are known (as the case of our experimental testbed). This insight is used to provide a heuristic that can be used to assess when the identification results are likely to have high or low accuracy.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

MPC solution for optimal load shifting for buildings with ON/OFF staged packaged units: Experimental demonstration, and lessons learned

Small and medium-sized commercial buildings (SMCB) are significant demand response resources, and it is important to develop grid-responsive control algorithms that exploit those resources and create financial benefits for building owners and HVAC service providers. Furthermore, unlike large-sized commercial buildings, there is an opportunity to have universally applicable control solutions for many SMCBs since those buildings have a consistent HVAC system configuration: SMCBs are commonly served by multiple-staged air conditioning units controlled by their own thermostats. Despite the demand response potential and scalability, however, very few control solutions are available for SMCBs. Typical model predictive control (MPC) and heuristic control approaches for cooling load shifting that lower thermostat setpoints before an electric price jump are suitable mainly for large-sized commercial buildings where a continuous capacity modulation is possible, e.g., via dampers in variable air volume terminal units. However, those approaches can cause undesired, high peaks for SMCBs due to the nature of ON/OFF unit staging and narrow thermostat deadbands. This could discourage the use of advanced grid-responsive controls for SMCBs due to the concern of high demand charges, and has to be resolved. This paper presents a MPC solution that overcomes this challenge. It has a hierarchical MPC structure where an upper level MPC is responsible for electrical load shifting in response to an electric price signal while a lower level MPC is responsible for coordinating compressor stages to eliminate unnecessary peaks and follows the setpoints determined by the upper level MPC. In this work, two one-month, comprehensive laboratory tests have been carried out to demonstrate load shifting and cost savings for the algorithm. Interesting trade-offs between energy efficiency and load flexibility were observed and are discussed, and lessons learned for applying MPCs for SMCBs are also presented.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Inverse reinforcement learning control for building energy management

Reinforcement learning (RL) based control is widely considered a promising approach in building automation and control as it has demonstrated the potential to deal with complex objectives in adjacent domains like robotics, autonomous vehicles, gaming applications, and advertisement recommendations. When applied to any environment, model-free RL learns to improve its control performance over time without requiring a control model, by receiving and then analyzing feedback from the building environment after each control action. Operational objectives are becoming increasingly complex through the simultaneous consideration of thermal comfort, carbon emissions, grid services, and indoor air quality. In this context, conventional rule-based control approaches are proving sub-optimal, mostly heuristic, and inadequate. The model-free and self learning nature of RL appears promising and attractive as it may address the scalability issues associated with advanced control approaches. However, it suffers from long training times and unstable control behavior during the early stages of its learning process, which makes it unsuitable to be applied directly to buildings. This paper addresses these issues using an inverse reinforcement learning approach (IRL), a technique utilized to learn the objective of a controller agent which is considered an expert in its respective domain. Here, we consider a rule-based control as the expert demonstrator. IRL is different from a direct imitation (i.e., direct mapping of states to actions) of control actions as it tries to find the underlying intent of an expert's policy, providing the controller with a better-generalized policy for unseen states or environments with slightly different dynamics. This approach propels the RL controller's policy to levels similar to or better than that of a rule-based policy before it starts learning by interacting with the building. This makes RL for building energy management applications more practical as it prevents the erratic and exploratory behavior in the initial training period, simultaneously speeding up the learning process when compared to applying an untrained RL agent directly to a building environment.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Large-scale simulation-based parametric analysis of an optimal precooling strategy for demand flexibility in a commercial office building

Achieving success with grid-interactive efficient buildings (GEBs) is closely tied to the utilization of flexible loads. A valuable strategy involves the implementation of precooling techniques before high-demand events, such as peak hours, by adjusting zone air temperature setpoints. This leads to a reduction in thermal loads and peak electricity demand during these times, as the building’s thermal mass stores and subsequently releases thermal energy. However, the effectiveness of the pre-cooling optimization is highly contingent on specific conditions such as building thermal properties, weather conditions, utility rate structure, HVAC equipment sizing, etc. Therefore, investigating the impacts of these condition-specific factors is crucial, especially when considering precooling strategies that utilize thermal mass in commercial buildings. In this paper, we first devised a novel heuristic control approach that incorporates parameterized optimal precooling thermostat schedules to enhance demand flexibility in a commercial office building. Subsequently, we conducted a thorough performance evaluation of this control strategy. Here, the optimal thermostat schedule was parameterized using three optimization variables: the precooling start time, the precooling end time, and the precooling temperature setpoint. Utilizing the DOE medium-sized office building as the virtual testbed, we showed that the parameterized schedule effectively approximates model predictive control and requires drastically reduced computational overhead. In addition, we investigated the impact of different influencing factors on the optimal precooling strategy. These factors include building thermal mass, outdoor air conditions, and energy price profiles. Using high-performance computing, we simulated a total of 225 scenarios, consisting of three levels of thermal mass, five typical outdoor air temperature profiles, and fifteen time-of-use price plans. The results demonstrate that optimal thermostat scheduling could save substantial energy cost in medium-sized office buildings with heavy thermal mass but with some energy penalty. Although the potential for cost savings is lower in buildings with low and medium thermal mass, the energy penalty remains consistent in all three thermal mass scenarios. The study also highlights the need to account for zone diversity and recognize that a one-size-fits-all-zone setpoint schedule may not be suitable for all zones and can lead to unnecessary energy wastage. Furthermore, the results highlight that while outdoor air conditions play a role in cost and energy performance, the cooling load exerts a more immediate and substantial influence on cost savings in precooling strategies. Although cost savings are comparable under certain conditions with the same cooling load, observed deviations in energy penalty indicate potential disparities in the efficiency of the HVAC system during the load-shifting process. In addition, the duration of peak pricing and the ratio between peak and off-peak times exhibit clear correlations with cost savings and energy consumption, aligning with intuitive expectations. These findings offer valuable insights for optimizing precooling strategies in office buildings.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

An optimization framework for the network design of advanced district thermal energy systems

In this work, a topology optimization framework for district thermal energy systems is presented. The framework seeks to address the questions, for a given district, "What is the best subset of buildings to connect to a district thermal energy system, and by what network should they be connected, to minimize life cycle cost?" A particle swarm optimization approach is validated to address the selection of the subset of buildings, and a graph theory-based heuristic is validated for selection of the network topology for any candidate subset of buildings. The framework is applied to a prototypical urban district for illustrative purposes. Additionally, modeling of prototypical districts revealed reductions in source energy use intensity for heating and cooling of 21-25% through the use of advanced district energy systems relative to code-compliant, building level systems. The framework identifies solutions with life cycle cost values 14% to 72% lower than that of base case scenarios based on conventional design approaches, depending on the base case scenario selected. Analysis of the search space indicates that topology optimization facilitates reductions in life cycle cost, source energy use intensity, and carbon emissions.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Energy R&D portfolio analysis based on climate change mitigation

The diverse nature and uncertain potential of the energy technologies that are or may be available to mitigate greenhouse gas emissions pose a challenge to policymakers trying to invest public funds in an optimal R&D portfolio. This paper discusses two analytical approaches to this challenge used to inform funding decisions related to the U.S. Department of Energy (DOE) applied energy R&D portfolio. The two approaches are distinguished by the constraints under which they were conducted: the need to provide an end-to-end portfolio analysis as input to internal DOE budgeting processes, but with limited time and subject to institutional constraints regarding important issues such as expert judgment. Because of these constraints, neither approach should be viewed as an attempt to push forward the state of the art in portfolio analysis in the abstract. Instead, they are an attempt to use more stylized, heuristic methods that can provide first-order insights in the DOE institutional context. Both approaches make use of advanced technology scenarios implemented in an integrated assessment modeling framework and then apply expert judgment regarding the likelihood of achieving associated R&D and commercialization goals. The approaches differ in the granularity of the scenarios used and in the definition of the benefits of technological advance: in one approach the benefits are defined as the cumulative emission reduction attributable to a particular technology; in the other approach benefits are defined as the cumulative cost reduction. In both approaches a return on investment (ROI) criterion is established based on benefits divided by federal R&D investment. The ROI is then used to build a first-order approximation of an optimal applied energy R&D investment portfolio. Although these methodologies have been used to inform an actual budget request, the results reflect only one input among many used in budget formulation. The results are therefore not representative of an official U.S. government or DOE funding recommendation but should instead be considered illustrative of the way in which methodologies such as these could be applied.

Climate change↗

A Reinforcement Learning Approach to Parameter Selection for Distributed Optimal Power Flow

With the increasing penetration of distributed energy resources, distributed optimization algorithms have attracted significant attention for power systems applications due to their potential for superior scalability, privacy, and robustness to a single point-of-failure. The Alternating Direction Method of Multipliers (ADMM) is a popular distributed optimization algorithm; however, its convergence performance is highly dependent on the selection of penalty parameters, which are usually chosen heuristically. In this work, we use reinforcement learning (RL) to develop an adaptive penalty parameter selection policy for alternating current optimal power flow (ACOPF) problem solved via ADMM with the goal of minimizing the number of iterations until convergence. We train our RL policy using deep Q-learning and show that this policy can result in significantly accelerated convergence (up to a 59% reduction in the number of iterations compared to existing, curvatureinformed penalty parameter selection methods). Furthermore, we show that our RL policy demonstrates promise for generalizability, performing well under unseen loading schemes as well as under unseen losses of lines and generators (up to a 50% reduction in iterations). This work thus provides a proof-of-concept for using RL for parameter selection in ADMM for power systems applications.

alternating current optimal power flow↗

Stochastic economic dispatch of wind power under uncertainty using clustering-based extreme scenarios

Operation of power systems with high penetrations of renewable energy sources requires tools for robust decision making under uncertainty. Stochastic economic dispatch and stochastic unit commitment are effective techniques for planning and operation under uncertainty, whose effectiveness depends on the cardinality and quality of the scenario set. Here, this article proposes a machine learning method using -means clustering for capturing relevant physical information from a large population of analog scenarios. Extreme scenario samples drawn from the clusters are used in a two-stage stochastic economic dispatch computation. The effectiveness of the proposed approach is assessed on a synthetic 200-bus system with a geographic footprint over Illinois, USA for four months from each season of WIND Toolkit data. The combination of -means clustering with importance sampling is shown to reduce the total operational cost by over 43% compared to sampling from populations based on heuristic clustering-based methods. Additionally, the variability in the mean cost is about 56% lower than the variability using Monte Carlo sampling. Moreover, the operational cost with the presented approach is shown to be close to the cost calculated based on a hindsight exact wind profile, signifying a highly accurate quantification of wind uncertainty by the presented -means clustering based sampling method.

17 WIND ENERGY↗

Dispatch optimization of electric thermal energy storage within System Advisor Model

A stand-alone electric thermal energy storage (ETES) system converts low-value electricity into heat using resistance heating elements. During periods of high-value electricity, an ETES system uses a thermodynamic power cycle to convert stored thermal energy back to electricity. These dispatchable systems derive value from their ability to store energy when prices are low and generate electricity when prices are favorable, i.e., energy arbitrage. Consequently, dispatch optimization of system operations, through maximizing revenue subject to system constraints, is essential to evaluate the economic value of a particular system design. While stand-alone ETES systems offer potential advantages as dispatchable grid storage technologies, there is a lack of a neutral third-party, publicly available, open-source model to evaluate the performance, dispatch, and financial viability of these systems. To address this problem, we have developed a techno-economic model for stand-alone ETES systems, within National Renewable Energy Laboratory's (NREL's) System Advisor Model (SAM). We implement a mixed-integer linear program to determine an ETES optimal operating schedule that maximizes electricity sales less maintenance costs caused by operation and cycling given temporal-varying grid electricity prices. Our contributions include a mixed-integer linear program for energy arbitrage of an ETES system, an ETES performance model through a publicly-available software (i.e., SAM), and an exercise of our model through case studies that compare ETES operational strategies and annual financial metrics. With our dispatch optimization model, we were able to improve revenue by 20% compared to a myopic heuristic while reducing the operational cost of the ETES system through decreases in cycle starts, cycles per day, and heater starts.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Automated design of an additive manufactured compact broadband antenna for plasma reflectometry

Broadband antennas operating in the gigahertz frequency range are regularly used for plasma reflectometry diagnostics. Due to a lack of space and unique diagnostic constraints, these antennas are often custom in design and frequency range. Recent advances in additive manufacturing of high temperature copper alloys allow for expanded freedom in design of these diagnostic antennas. In this work, a heuristic simulated annealing algorithm is used alongside 3-D finite element simulation to automate the design of a double ridged rectangular horn antenna for a reflectometry diagnostic on the DIII-D tokamak. Optimization of antenna performance given the design constraints results in a compact broadband (6-20+ GHz) antenna design. Measured transmission from the additively manufactured antenna matches simulation within reasonable error, and experimental plasma electron density profiles from the DIII-D high-field side scrape-off layer are shown.

Additive manufacturing↗

Energy-efficient mapping of large-scale workflows under deadline constraints in big data computing systems

Large-scale workflows for big data analytics have become a main consumer of energy in data centers where moldable parallel computing models such as MapReduce are widely applied to meet high computational demands with time-varying computing resources. The granularity of task partitioning in each moldable job of such big data workflows has a significant impact on energy efficiency, which remains largely unexplored. In this paper, we analyze the properties of moldable jobs and formulate a workflow mapping problem to minimize the dynamic energy consumption of a given workflow request under a deadline constraint in big data systems. Since this problem is strongly NP-hard, we design a fully polynomial-time approximation scheme (FPTAS) for a special case with a pipeline-structured workflow on a homogeneous cluster and a heuristic for the generalized problem with an arbitrary workflow on a heterogeneous cluster. The performance superiority of the proposed solution in terms of dynamic energy saving and deadline missing rate is illustrated by extensive simulation results in comparison with existing algorithms, and further validated by real-life workflow implementation and experimental results in Hadoop/YARN systems.

97 MATHEMATICS AND COMPUTING↗

ULAMA: A Utilization-Aware Matching Approach for robust on-demand spatial service brokers

This article investigates an on-demand spatial service broker for suggesting service provider propositions and the corresponding time of service to mobile consumers while meeting the consumer’s maximum travel time and wait time constraints. The goal of the broker is to maximize the number of matched requests while also keeping the “eco-system” functioning and robust against variations in the supply–demand ratio by engaging as many service providers as possible and maintaining providers’ participation to ensure that providers do not drop out. This problem is important because of its many related societal applications in the on-demand and sharing economy (e.g., on-demand ride-hailing services, on-demand food delivery, etc.). Challenges of this problem include the need to satisfy many conflicting requirements of the broker, consumers and service providers and the high computational complexity since solving the problem for a number of available consumers at any time instant is NP-hard. Related work in spatial crowdsourcing and ridesharing has mainly focused on maximizing the number of matched requests and minimizing travel cost, but did not consider the importance of maintaining provider engagement and balancing provider utilization, which could become a priority when the available supply exceeds the demand. In this paper, we propose a Utilization-Aware Matching Approach (ULAMA) which employs novel provider-centric heuristics for balancing the utilization of providers, and a consumer-priority-based greedy matching algorithm that prioritizes consumers for maximizing the number of matched requests. Experimental results show that our proposed approach outperforms related work by achieving the lowest variance in provider utilization while matching all available providers even when supply greatly exceeds demand. Our approach also achieved a larger number of matched requests, particularly when supply exceeds demand and also when both supply and demand are balanced.

97 MATHEMATICS AND COMPUTING↗

QoS-aware edge AI placement and scheduling with multiple implementations in FaaS-based edge computing

Resource constraints on the computing continuum require that we make smart decisions for serving AI-based services at the network edge. AI-based services typically have multiple implementations (e.g., image classification implementations include SqueezeNet, DenseNet, and others) with varying trade-offs (e.g., latency and accuracy). The question then is how should AI-based services be placed across Function-as-a-Service (FaaS) based edge computing systems in order to maximize total Quality-of-Service (QoS). To address this question, we propose a problem that jointly aims to solve (i) edge AI service placement and (ii) request scheduling. These are done across two time-scales (one for placement and one for scheduling). Here we first cast the problem as an integer linear program. We then decompose the problem into separate placement and scheduling subproblems and prove that both are NP-hard. We then propose a novel placement algorithm that places services while considering device-to-device communication across edge clouds to offload requests to one another. Our results show that the proposed placement algorithm is able to outperform a state-of-the-art placement algorithm for AI-based services, and other baseline heuristics, with regard to maximizing total QoS. Additionally, we present a federated learning-based framework, FLIES, to predict the future incoming service requests and their QoS requirements. Our results also show that our FLIES algorithm is able to outperform a standard decentralized learning baseline for predicting incoming requests and show comparable predictive performance when compared to centralized training.

97 MATHEMATICS AND COMPUTING↗

Alternative mixed integer linear programming optimization for joint job scheduling and data allocation in grid computing

This paper presents a novel approach to the joint optimization of job scheduling and data allocation in grid computing environments. We formulate this joint optimization problem as a mixed integer quadratically constrained program. To tackle the nonlinearity in the constraint, we alternatively fix a subset of decision variables and optimize the remaining ones via Mixed Integer Linear Programming (MILP). We solve the MILP problem at each iteration via an off-the-shelf MILP solver. Our experimental results show that our method significantly outperforms existing heuristic methods, employing either independent optimization or joint optimization strategies. We have also verified the generalization ability of our method over grid environments with various sizes and its high robustness to the algorithm setting.

97 MATHEMATICS AND COMPUTING↗