مدلسازی برای ذخیره سازی و مدیریت تقاضا در شبکه های توزیع برق

Storage devices and demand control may constitute beneficial tools to optimize electricity generation with a large share of intermittent resources through inter-temporal substitution of load. This paper quantifies the related cost reductions in a simulation model of a simplified stylized medium-voltage grid (10 kV) under uncertain demand and wind output. Benders Decomposition Method is applied to create a two-stage stochastic optimization program. The model informs an optimal investment sizing decision as regards specific ‘smart’ applications such as storage facilities and meters enabling load control. Model results indicate that central storage facilities are a more promising option for generation cost reductions as compared to demand management. Grid extensions are not appropriate in any of the scenarios. A sensitivity analysis is applied with respect to the market penetration of uncoordinated Plug-In Electric Vehicles which are found to strongly encourage investment into load control equipment for ‘smart‘ charging and slightly improve the case for central storage devices

Since electricity demand and the availability of output from Renewable Energy Sources (RES) are intermittent by nature, system operators have to resort to relatively costly measures such as reserve energy to maintain system stability. Back-up capacities are set to become more relevant with increasing shares of RES penetration. In this context, storage devices serve to store excessive electricity generation and feed-in missing energy in times of need. An alternative concept of better aligning demand and supply of electricity through two-way digital communication technology is commonly referred to as ‘smart metering’. Measures to manage demand with the help of smart meters include demand response and direct load control. Recent legislation obliges German grid operators and utilities to install smart metering systems in new and refurbished dwellings. While legislative pressure spurs investment in smart metering, it may imply a negative effect on investment incentives in storage. This paper scrutinizes load control and storage facilities as potential concurrent options targeting at electricity generation cost reductions and it quantifies possible substitution effects. Because of their common purpose, direct load control and centralised storage are two competing or possibly complementary solutions from the perspective of a vertically integrated power distribution system operator and utility. Moreover, it is tested whether storage and load control could alleviate the need for grid reinforcements by avoiding capacity shortages. The idea is that avoided shortage adds value to storage or DSM devices because of capacity upgrade deferral and added electricity sales [1]. Additionally to these issues, a methodological purpose of this paper is to demonstrate how stochastic optimization and Benders Decomposition Method can be sensibly applied to analyze and compare investment options in a power distribution system setting. The focus lies on short-term uncertainties and their impact on investment decisions. There exists a broad range of literature dealing with storage sizing decisions. Refs. [2], [3], [4], [5] and [6] perform numerical optimizations in a deterministic setting. Applications of stochastic patterns of generation and demand can be found in [7], [8], [9] and [10]. Tan et al. [10] present a stochastic optimization model of battery sizing for demand management with emphasis on outage probabilities which is not dealt with in this paper. Roy et al. [11] apply stochastic wind generation patterns to a wind-battery system sizing model with deterministic demand. Ref. [12] do likewise with Plug-in Electric Vehicles (EV) as storage facilities. The combination of intermittency of renewable resources and demand-side-management (DSM) is addressed in [13] and [14]. Concerning demand-side management (DSM), numerous research publications were found on investment decisions into DSM. Ki Lee et al. [15] assess investment into demand management systems for heating in a national case study for Korea. Paulus and Borggrefe [16] adopt a system-wide perspective of investment in DSM in a case study for Germany with focus on industrial consumers. Manfren et al. [17] deal with distributed generation planning, but avoid making any investment analysis. Neenan and Hemphill [18] investigate investment from a societal perspective while [19] and [20] find that investment into DSM appliances might not be all that profitable in general. It is intended to further investigate this claim in the present analysis. This paper’s contribution is unique in that no study explicitly compares the cost saving potential of storage and DSM in a comprehensive model including grid representation, endogenous investment and factors of uncertainty. Whilst an 11 kV distribution network representation in combination with a benefit analysis for storage and demand response measures can be found in [21], the present work complements their analysis by adding endogeneity to the investment into storage devices and DSM appliances as well as uncertainty of demand and wind generation. A further contribution consists in the application of Benders Decomposition Method to the stochastic program. Decomposition methods can be applied to numerous bi-level optimization problems in the energy sector, such as unit-commitment or capacity expansion. To the author’s best knowledge, an application to evaluating storage and DSM infrastructure investment is unprecedented. The article is divided into a descriptive part, including the methodology and model description, an explanation of parameters and scenarios applied. Subsequently, results are outlined, discussed and final conclusions are drawn.

This paper presents a DC load flow model applied to investment in storage and DSM facilities in a stylized medium-voltage grid. The model incorporates uncertainty in demand and wind output and uses Benders Decomposition to distinguish the investment choices from operative optimizations. It is shown how Benders Decomposition Method can be meaningfully applied to a small-scale investment problem in a network-constrained industry. The model is capable of reflecting multiple formats of short-term uncertainties in system constraints at the operational dispatch stage. The model results indicate that grid reinforcements at 10 kV level are not necessary in any scenario. Capacity utilization rates do not hit the 60% bound, which implies there is little harm to system stability. Results suggest that storage devices are beneficial at capacity cost of up to 850 EUR/kW h under the stipulated conditions. This implies that relatively expensive storage technologies such as Nickel–Cadmium and Nickel-metal hydride storage are profitable at current cost. Flywheels and large-scale capacitors are not competitive unless cost is reduced to 25% of 2011 cost. DSM is not beneficial in any scenario, particularly in the deterministic model. Investment is beneficial up to an all-inclusive cost of roughly 200 EUR per consumer. This break-even point (tolerance threshold) boosts when consumers own EV, implying that EV strongly encourage investment into load control systems. The finding reflects the actual fact that most EV are sold along with advanced (‘smart’) metering systems. As a logical consequence, it is found that investment into storage is likely to crowd out investment into DSM appliances in the model setting. Since both options are direct alternatives for energy management, ‘smart meters’ seem to be of little economic value to the system operator in the absence of EV. Unless governments strongly encourage DSM through obligations (beyond current obligations) and financial incentives or the promotion of EV, storage facilities are the better option for a vertically integrated distribution system operator facing the conditions of this model. The present paper aimed at modeling conditions that would be representative for a section of a stylized distribution system in Germany. It could be shown, that the stochastic model produces more efficient solutions than its deterministic counterpart. The cost of disregarding uncertainty lies at 0.5–5% of total generation cost. The analysis demonstrates that a stochastic treatment of wind and demand patterns significantly augments the case for the use of storage. The break-even point for investment decisions into storage increases from 350 to 850 EUR/kW h when uncertainty of wind and demand are taken into account. Hence, the deterministic model leads to considerable under-investment into storage. All in all, the results are highly sensitive to the assumed investment cost for storage and load management devices. EV are another cause for variations, yet, to a lesser extent. The calculations indicate that the value of storage strongly varies with the intermittency of wind output. The value of DSM is less sensitive to wind but more sensitive to EV penetration. There are a number of conceptual caveats to the analysis which constitute areas for improvement. Energy saving through demand response is entirely factored out. The model may therefore underestimate the value of DSM to a minor extent. Furthermore, the investment cost for batteries is calculated on a diurnal basis with a fixed number of cycles per day. Fixing the cycles is a necessary step to obtain an exogenous cost figure but somewhat arguable since the cycles are endogenously determined in the model. Another drawback of this model is that some potential business cases of batteries and DSM are not included. Besides peak load reductions and network reinforcement deferral, [21] point to other benefits of using storage devices. For instance, balancing markets as potential business field for batteries are not included in the present model. Other shortcomings are the stylized grid configuration and the absence of ramping constraints for storage, which can be included in a further model of larger size. An application to a grid of larger size is planned for a subsequent paper.