Jacint Szabo

Aggregation of energetic flexibility using zonotopes

The importance of energetic flexibility of distributed energy resources grows with the share of renewable generation in the power grid. However, the quantitative description and aggregation of flexible resources is challenging. This work proposes the use of zonotopes, a subclass of polytopes, to approximate flexibility. It is shown how optimal zonotopic approximations of flexibility can be computed efficiently for different objectives, and that the aggregation of those sets is tractable with regard to memory and computational complexity for long planning horizons and large populations of systems. In addition, we describe synergistic behavior exhibited by the aggregation of flexibility and illustrate that zonotopes can partly capture these synergy effects.

Aggregation and Disaggregation of Energetic Flexibility from Distributed Energy Resources

A variety of energy resources has been identified as being flexible in their electric energy consumption or generation. This energetic flexibility can be used for various purposes such as minimizing energy procurement costs or providing ancillary services to power grids. To fully leverage the flexibility available from distributed small-scale resources, their flexibility must be quantified and aggregated. This paper introduces a generic and scalable approach for flexible energy systems to quantitatively describe and price their flexibility based on zonotopic sets. The description allows aggregators to efficiently aggregate the flexibility of large numbers of systems and to make control and market decisions on the aggregate level. In addition, an algorithm is presented that distributes aggregate-level control decisions among the individual systems of the population in an economically fair and computationally efficient way. The algorithm is applied to the problem of disaggregating reference schedules resulting from day-ahead energy markets.

Arrival time dependent routing policies in public transport

Abstract We present a routing system that considers uncertainties, which are prevalent in any real transport system. Given desired departure or arrival times and a utility function representing the traveller’s preferences, our method computes not just a single path through the network, but a more sophisticated and adaptive journey plan called routing policy. For each stop and time instance, a policy specifies the list of services that the passenger is recommended to take. We show that the problem of finding an optimal policy is NP-hard. We also give a polynomial-time algorithm for a relaxation of the problem when the number of recommended services is limited at each stop and time. A computational case study for the public transport network of Budapest shows that the obtained routing policies can lead to substantial travel time savings compared to deterministic plans, and that considering multiple service policies leads to an improvement compared to previous solutions using single-service policies.

Predicting DRAM reliability in the field with machine learning

Uncorrectable errors in dynamic random access memory (DRAM) are a common form of hardware failure in server clusters. Failures are costly both in terms of hardware replacement costs and service disruption. While a large body of work exists on analyzing DRAM reliability in large production clusters, little has been reported on the automatic prediction of such errors ahead of time. In this paper, we present a highly accurate predictive model, based on daily event logs and sensor measurements, in a large fleet of commodity servers going back to 2014. By correlating correctable errors with sensor metrics, we can use ensemble machine learning techniques to predict uncorrectable errors weeks in advance.n In addition, we show how such models can be applied in the wild and consumed by customer support teams. Our goal is to minimize false positives, as healthy DRAMs should not be replaced, while accounting for common limitations, such as missing data points and rare occurences of uncorrectable errors.