Contingency-Constrained Unit Commitment with High Penetration of Intermittent Renewables

可靠性是电力系统的一个主要问题，工程师们一直在努力在各种不确定因素下保持灯光亮起，包括发电机或输电线路突然故障造成的意外情况。北美电力可靠性公司(NERC)N-1规则要求这种单一故障不应导致系统不可行，该规则已嵌入机组组合过程的某些版本中。由此产生的应急约束机组组合(CCUC)极其复杂，而风能或太阳能等间歇性可再生能源的急剧增加现在又加剧了这一问题。例如，在西班牙，短短几个小时内风力发电量下降相当于关闭四个核电机组的情况并不少见，这给系统运营商带来了重大挑战。因此，在间歇性可再生能源存在的情况下考虑意外情况是一个关键而紧迫的问题。与ISO-NE、南加州爱迪生和阿尔斯通电网合作，该提案针对间歇性可再生能源渗透率较高的CCUC。考虑到不同类型的不确定性及其大小、大量的意外事件、混合整数优化的组合性质以及电力系统的广泛性，这一问题是具有挑战性的。我们的重点将放在区间CCUC的新配方和先进的优化方法上。这项拟议的研究将对管理其他不确定性，如N-2意外事件，以及其他复杂问题，如微电网能源优化，产生更广泛的影响。对于这些问题，与不确定的天气、电力需求和微电网相关的间歇性可再生能源相关的困难可以用区间公式来建模。由此产生的混合整数线性优化问题可以从我们先进的优化方法中受益。此外，这项研究将通过课程、研讨会、出版物和在线材料来教育研究生、本科生和专业人员。也会通过他们的老师让高中生受益吗？参与了康涅狄格州大学的达芬奇夏季项目。将积极鼓励代表性不足群体的参与。具体而言，任务1是制定考虑N-1个应急事件和间歇性可再生能源的能源和储备共同优化的CCUC。新的思想在于使用区间来描述基于发电转移因数(GSFS)的输电事故，而不是一次分析一个事故。线流是净节点注入的加权和，其中权重是GSFS。当输电线路跳闸时，GSFS将改变。因此，各种传输意外情况可以通过改变GSF的值来描述，并且可以通过间隔来建模。通过基于识别和消除冗余的应急传输约束来缩小GSF的范围，以及通过改进区间算法，例如通过使用仿射算法，将减轻该建模的保守性。发电机意外事故和间歇性可再生能源随后将纳入区间优化框架。为了有效地解决这个问题，任务2正在开发新一代优化方法。关键是将我们最新的代理拉格朗日松弛(SLR)和分支切割(SLR)进行协同集成，SLR利用可分离性，分支和切割利用线性。虽然SLR需要迭代，但通过新颖地使用子问题凸壳的不变性质来迭代，总的CPU时间和解的质量将得到显著改善。此外，我们还将探索我们的方法的并行实施。该方法将使用来自ISO-NE和CAISO的真实数据集进行测试。

Reliability is a major concern for power systems, and engineers have been striving hard to keep lights on under all kinds of uncertainty, including contingencies caused by a sudden failure of a generator or a transmission line. The North American Electric Reliability Corporation (NERC) N-1 rule requires that no such single failure should result in system infeasibility, and this rule has been embedded in certain versions of the unit commitment process. The resulting contingency-constrained unit commitment(CCUC) is extremely complex, and the issue is now compounded by the drastic increase in intermittent renewables such as wind or solar. In Spain, for example, it is not uncommon that a drop of wind generation within just a few hours is equivalent to the shutdown of four nuclear units, and this has presented major challenges to system operators. Consideration of contingencies in the presence of intermittent renewables is thus a critical and urgent issue. In collaboration with ISO-NE, Southern California Edison and Alstom Grid, this proposal is on CCUC with high penetration of intermittent renewables. The problem is challenging in view of different types of uncertainties and their magnitudes, the large number of contingencies, the combinatorial nature of mixed-integer optimization, and the largeness of power systems. Our focus will be on novel formulations of interval CCUC, and on advanced optimization methods. The proposed research will have the broader impacts on managing other uncertainties, e.g., N-2 contingencies, and on other complex problems such as microgrid energy optimization. For these problems, difficulties associated with uncertain weather, electricity demand and microgrid-associated intermittent renewables can be modeled by the interval formulation. The resulting mixed-integer linear optimization problems can then benefit from our advanced optimization methods. Furthermore, the research will educate graduate and undergraduate students and professionals through courses, seminars, publications and online materials. It will also benefit high school students through their teachers? involvement in the daVinci summer project at UConn. The participations of under-represented groups will be actively sought.Specifically, Task 1 is on formulating CCUC for the co-optimization of energy and reserve considering N-1 contingencies and intermittent renewables. The novel idea lies in using intervals to describe transmission contingencies based on generation shift factors (GSFs), as opposed to analyzing contingencies one at a time. A line flow is a weighted sum of net nodal injections, where weights are GSFs. When a transmission line is tripped, GSFs will be changed. Various transmission contingencies can thus be described by varying values of GSFs, and be modeled by intervals. Conservativeness of this modeling will be alleviated through shrinking ranges of GSFs based on identifying and eliminating redundant contingency transmission constraints, and through improving interval arithmetic, e.g., by using affine arithmetic. Generator contingencies and intermittent renewables will then be incorporated within the interval optimization framework. To effectively solve the problem, Task 2 is on developing a new generation of optimization methods. The key is to synergistically integrate our latest surrogate Lagrangian relaxation(SLR), which exploits separability, with branch-and-cut, which exploits linearity. While SLR requires iterations, by novel exploitation of the invariant nature of subproblem convex hulls across iterations, the overall CPU time and solution quality will be drastically improved. In addition, parallel implementation of our approach will be explored. The method will be tested using realistic data sets from ISO-NE and CAISO.