Stephen Haben

A Peak Reduction Scheduling Algorithm for Storage Devices on the Low Voltage Network

Reinforcing the Low Voltage (LV) distribution network will become essential to ensure it remains within its operating constraints as demand on the network increases. The deployment of energy storage in the distribution network provides an alternative to conventional reinforcement. This paper presents a control methodology for energy storage to reduce peak demand in a distribution network based on day-ahead demand forecasts and historical demand data. The control methodology pre-processes the forecast data prior to a planning phase to build in resilience to the inevitable errors between the forecasted and actual demand. The algorithm uses no real time adjustment so has an economical advantage over traditional storage control algorithms. Results show that peak demand on a single phase of a feeder can be reduced even when there are differences between the forecasted and the actual demand. In particular, results are presented that demonstrate when the algorithm is applied to a large number of single phase demand aggregations that it is possible to identify which of these aggregations are the most suitable candidates for the control methodology.

Conditioning of incremental variational data assimilation, with application to the Met Office system

Implementations of incremental variational data assimilation require the iterative minimization of a series of linear least-squares cost functions. The accuracy and speed with which these linear minimization problems can be solved is determined by the condition number of the Hessian of the problem. In this study, we examine how different components of the assimilation system influence this condition number. Theoretical bounds on the condition number for a single parameter system are presented and used to predict how the condition number is affected by the observation distribution and accuracy and by the specified lengthscales in the background error covariance matrix. The theoretical results are verified in the Met Office variational data assimilation system, using both pseudo-observations and real data.

A hybrid model of kernel density estimation and quantile regression for GEFCom2014 probabilistic load forecasting

We present a model for generating probabilistic forecasts that combines the kernel density estimation (KDE) and quantile regression techniques, as part of the probabilistic load forecasting track of the Global Energy Forecasting Competition 2014. Initially, the KDE method is implemented with a time-decay parameter, but we later improve this method by conditioning on the temperature or period of the week variables in order to provide more accurate forecasts. Secondly, we develop a simple but effective quantile regression forecast. The novel aspects of our methodology are two-fold. First, we introduce symmetry into the time-decay parameter of the kernel density estimation based forecast. Second, we combine three probabilistic forecasts with different weights for different periods of the month.

A genetic algorithm approach for modelling low voltage network demands

Abstract Distribution network operators (DNOs) are increasingly concerned about the impact of low carbon technologies on the low voltage (LV) networks. More advanced metering infrastructures provide numerous opportunities for more accurate load flow analysis of the LV networks. However, such data may not be readily available for DNOs and in any case is likely to be expensive. Modelling tools are required which can provide realistic, yet accurate, load profiles as input for a network modelling tool, without needing access to large amounts of monitored customer data. In this paper we outline some simple methods for accurately modelling a large number of unmonitored residential customers at the LV level. We do this by a process we call buddying, which models unmonitored customers by assigning them load profiles from a limited sample of monitored customers who have smart meters. Hence the presented method requires access to only a relatively small amount of domestic customers’ data. The method is efficiently optimised using a genetic algorithm to minimise a weighted cost function between matching the substation data and the individual mean daily demands. Hence we can show the effectiveness of substation monitoring in LV network modelling. Using real LV network modelling, we show that our methods perform significantly better than a comparative Monte Carlo approach, and provide a description of the peak demand behaviour.

Modelling the Electricity Consumption of Small to Medium Enterprises

Estimating the demand on the low voltage network is essential for the distribution network operator (DNO), who is interested in managing and planning the network. Such concerns are particularly relevant as the UK moves towards a low carbon economy, and the electrification of heating and transport. Furthermore, small to medium enterprises (SMEs) contribute a significant proportion to network demand but are often overlooked. The smart meter roll out will provide greater visibility of the network, but such data may not be readily available to the DNOs. The question arises whether useful information about customer demand can be discerned from limited access to smart meter data? We analyse smart meter data from 196 SMEs so that one may create an energy demand profile based on information which is available without a smart meter. The profile itself comprises of simply two estimates, one for operational power and another for non-operational power. We further improve the profile by clustering the SMEs using a simple Gaussian mixture model. In both cases, the average difference between the actual and predicted operational/non-operational power is less than 0.15 kWh, and clustering reduces the range around this difference. The methods presented here out perform the flat profile (akin to current methods).

The conditioning of least squares problems in variational data assimilation

Summary In variational data assimilation a least-squares objective function is minimised to obtain the most likely state of a dynamical system. This objective function combines observation and prior (or background) data weighted by their respective error statistics. In numerical weather prediction, data assimilation is used to estimate the current atmospheric state, which then serves as an initial condition for a forecast. New developments in the treatment of observation uncertainties have recently been shown to cause convergence problems for this least-squares minimisation. This is important for operational numerical weather prediction centres due to the time constraints of producing regular forecasts. The condition number of the Hessian of the objective function can be used as a proxy to investigate the speed of convergence of the least-squares minimisation. In this paper we develop novel theoretical bounds on the condition number of the Hessian. These new bounds depend on the minimum eigenvalue of the observation error covariance matrix and the ratio of background error variance to observation error variance. Numerical tests in a linear setting show that the location of observation measurements has an important effect on the condition number of the Hessian. We identify that the conditioning of the problem is related to the complex interactions between observation error covariance and background error covariance matrices. Increased understanding of the role of each constituent matrix in the conditioning of the Hessian will prove useful for informing the choice of correlated observation error covariance matrix and observation location, particularly for practical applications.

Distribution Network Demand and Its Uncertainty

This chapter presents some advanced tools for low voltage (LV) network demand simulation. Such methods will be required to help distribution network operators (DNOs) cope with the increased uptake of low carbon technologies and localised sources of generation. This will enable DNOs to manage the current network, simulate the effect of various scenarios and run load flow analysis. In order to implement such analysis requires high resolution smart meter data for the various customers connected to the network. However, only small amounts of individual smart meter data will be available and such data could be expensive. In likelihood, smart meter data is only going to be freely available at the aggregate level. Hence, in general, to implement LV network tools, customer loads will need to be simulated based on the assumption of limited amounts of monitored data. In addition, due to the high volatility of LV electric distribution networks, demand uncertainty must also be captured within a simulation tool. In this chapter, a number of methods are described for simulating demand on low voltage feeders which rely only on relatively small samples of smart meter data and monitoring. Firstly, a method called ‘buddying’ is described for assigning realistic profiles to unmonitored customers by buddying them to a customer who is monitored. Secondly, a number of methods are presented for capturing the uncertainty on the network. Finally the uncertainty models are incorporated into the buddying method and implemented in a load flow analysis tool on a number of real feeders. Both the buddying and the uncertainty estimation are presented for two different cases based on whether LV substation monitoring is present or not. This illustrates the different impacts of monitoring availability on the modelling tools. This chapter demonstrates the presented methods on a large range of real LV feeders.

Smart Storage Scheduling and Forecasting for Peak Reduction on Low-Voltage Feeders

The transition to a low carbon economy will likely bring new challenges to the distribution networks, which could face increased demands due to low-carbon technologies and new behavioural trends. A traditional solution to increased demand is network reinforcement through asset replacement, but this could be costly and disruptive. Smart algorithms combined with modern technologies can lead to inexpensive alternatives. In particular, battery storage devices with smart control algorithms can assist in load peak reduction. The control algorithms aim to schedule the battery to charge at times of low demand and discharge, feeding the network, at times of high load. This study analyses two scheduling algorithms, model predictive control (MPC) and fixed day-ahead scheduler (FDS), comparing against a set-point control (SPC) benchmark. The forecasts presented here cover a wide range of techniques, from traditional linear regression forecasts to machine learning methods. The results demonstrate that the forecasting and control methods need to be selected for each feeder taking into account the demand characteristics, whilst MPC tends to outperform the FDS on feeders with higher daily demand. This chapter contributes in two main directions: (i) several forecasting methods are considered and compared and (ii) new energy storage control algorithm, MPC with half-hourly updated (rolling) forecasts designed for low voltage network application, is introduced, analysed and compared.