Load Shedding and Demand Side Management Enhancements to improve the Security of a National Electrical System• Antonio Capozza1, Christian D’Adamo2, Giuseppe Mauri1, Antonio Pievatolo3 1 CESI SpA, Via Rubattino 54, 20134 Milano tel. +39-0221251fax +39-0221255520 Email:giuseppe.mauri@cesi.it 2 Enel Distribuzione – via Ombrone, 2 – 00198 Roma 3 CNR-IMATI, via Bassini 15, 20133 Milano Recent blackouts, which happened in many countries worldwide [Man2004], brought to the attention of public opinion the importance of the security and reliability of the electric service. In a liberalised and fragmented market, with many different operators involved, ancillary services are provided by several actors. Electric utilities contribute to the security of the electrical system, according to guidelines and rules given by Independent System Operators, assuring load shedding in emergency cases. The widespread of new Automated Meter Reading (AMR) systems [Pil2003], will allow to better manage the electric load and could contribute in the improvement of security of electric systems reducing or shifting the peaks [Mau2003; Cap2000]. A main strategy currently implemented to assure security and reliability during emergency situation consists in load shedding both manual and automatic by using frequency relays installed in HV/MV substation [Sab2004]. These relays, sensible both to frequency thresholds and frequency variations, allow to sudden shed the load in case of a system collapse. Other systems consist in industrial load shedding (for example Industrial Customers with interruptible power contracts) or rotational power cuts for diffused customers. Short term solutions investigated to improve the current situation consist in new load shedding devices (with both under-frequency and under-voltage thresholds) and in the contribution of distributed generation to under-frequency transients. New load shedding relays with under voltage and voltage variation thresholds could improve the effectiveness of the load reduction at the interface of interconnected networks. By using new criteria in grid connection of distributed generation (DG) it will be possible to make DG contribution to sub-frequency transients, improving the stability of electric system. However all these systems and load shedding devices cause interruption to final users. Long term solutions aim to use features of new Electronic Meters and Automated Meter Reading (AMR) systems to manage peak demand by using both the tariff leverage and focused load reductions. Of course, maximum selectivity can be assured in power cuts or limitations. In this case, Demand Side Management (DSM) offers interesting perspectives even to improve the electric system security. Availability of new powerful electronic meters [Pil2003] and Power Line Communication networks of the latest generation [Cap et al 2003] brought Automated Meter Reading (AMR) systems to become the main support to services that electricity utilities need to offer [Mau2004]. These services encompass: • Providing Load shedding in emergency situation, safeguarding a few typologies of small and medium consumers • Making the participation of small and medium consumers to the market possible, through consumers aggregators (typically distribution utilities) • Improving operation and development of distribution grids Demand management requires to exchange information between utilities, meters and load control devices (when available at customer side) as shown in Figure 1. The content of messages may range from energy prices and meter readings, to complex data structures and commands in case of customer deploying advanced load control systems or, even, in case that third party companies are responsible for the energy management at user premises. The paper reviews AMR system architectures and analyses impact of new requirements on key components. Remote control centres and concentrators of MV/LV substations require the greatest functional enrichment, which will be discussed in depth. AMR system functions need only few adaptations. Several energy consumption values must be saved in meters at least for one day to allow their deferred transmission to the AMR control centre. New specific tasks must be assigned to concentrators in MV/LV substations, which have to collect (for current and next day), the type of tariff and control strategy to be adopted for each user involved in the DSM programme. For current day, concentrators should monitor load at substation level, verify risk of reaching threshold limits, identify reasons of deviations and, if necessary, react on the basis of the selected load control strategy. Concentrators can also be involved by the distribution control centre in monitoring the behaviour of MV/LV transformers. Main technologies used in AMR communication systems are PLC and GSM, which have strong economic justifications. The evolution of GSM to GPRS and EDGE can increase speed and allow to send powerful "broadcast" and "multicast" messages, but can introduce weakness that increase the vulnerability level. The paper investigates such solutions. Both load shedding at HV/MV substations and demand management through AMR functions can benefit from statistical models for load profile fitting and forecasting. Such models employ a wide range of different techniques (see • This work has been supported by MAP (Italian Ministry for Productive Activities) in the frame of Public Interest Energy Research Project named "Ricerca di Sistema" (MAP Decree of 28 February 2003). PUBBLICATO A5018612 (PAD – 635748)

2/3 [Mog1989] for a review and [Cav1998] for an application). Using a combination of regression and times series models, it is possible to obtain accurate short-term predictions of the load at any load point. The inclusion of a switching term (in the fashion of [Pel1996]) takes into account load-shedding actions. Then, from historical data, it is possible to estimate the parameters of the model, which can be used to predict the residual power drawn by the uninterruptible customers connected to a given load point when an emergency load shedding action takes place. Considering also previous experiences in the fields of: • Demand evolution [Cia2004] • Demand dependence on meteo-climatic conditions (e.g. interaction between heating/cooling plants and building) [Cap2004a] • Effective and practiced measures for implementation of DSM programmes (e.g. Computer Integrated Building, energy storage) [Cap2004b] the interest was focused on DMS using AMR functions. As a result of these studies, the predicted load was found to obey an equation similar to the following one: . 011 0 tit r =i i q =i ii p =i i
+d(t)+w(t)+Tc+nb+i)l(ta+a=l(t) -


– The equation says that the load at the end of hour t is a function: of the loads during the previous p hours; of the number ni of meters where the maximum allowable load has been reduced by, say, one kW (with one different bi parameter for each customer category, indexed by i); of the current temperature and of the temperature during the previous q hours (taking into account the thermal inertia of the buildings); of other weather variables w(t); of the time of day d(t). Typically, different parameters will be estimated for weekdays, holidays and weekends. The model, after estimating the parameters with historical data, is used to devise a load control strategy through the different predicted values obtained by varying the ni’s. Such a procedure is obviously necessary because of the uncertainty about the effect of a load limitation on a single customer, which can be null if his actual load is below the new threshold. In a similar fashion tariff profiles [Ram1995] can be included via the fraction of customers of each given category with a given profile: by following the evolution of the load profile at any given load point (such as an MV/LV substation) as the customers naturally change their tariff profiles, it will be possible to predict also the effect of customers changing their tariff profile for the next day as they participate in the energy market. Bibliography: [Man2004] A. Manzoni “Elementi comuni tra i blackout: 14 Agosto US-Canada, 28 settembre Svizzara-italia” AEI Vol. 91 Jul-Aug 2004 [Sab2004] C. Sabelli “Il piano per la sicurezza del sistema elettrico” AEI Vol. 91 Jul-Aug 2004 [Mau2004] G. Mauri “Applicabilità delle tecnologie attuali e future al controllo della domanda” CESI- A4501811, Ricerca di Sistema Seconda Fase, 2004 [Mau2003] G. Mauri “Analisi delle tecnologie dei contatori digitali e delle comunicazioni associate” CESI- A3042272 11/12/2003, Ricerca di Sistema Seconda Fase, 2003 [Cap2000] A. Capozza “Disamina delle metodologie per le valutazioni tecnico economiche di Gestione della Domanda (Demand Side Management) nell’industria elettrica; CESI- A0/021337. Ricerca di Sistema Prima Fase, 2000 [Dap et al. 2004] D.Apice et al. “Applicazione di Tecnologie Power Line e la Flessibilizzazione della Domanda” XXI riunione annuale del Gruppo di Misure Elettriche ed Elettroniche dell’AEIT: GMEE 2004, 16-18 Settembre, Crema, Italy, 2004 [Pil2003] Francesco Pilato. “Sistemi di controllo reti BT e Terminali di consegna e misura ”ACEST, Milano. Convegno su “La telelettura dei contatori d’utenza: i sistemi AMR come strumento per incrementare l’efficienza ed il business delle aziende di servizi di rete”, Milano 23 Settembre 2003 [Cap et al 2003] L.Capetta, F.Cesena; G.Mauri, R.Napolitano “Stato attuale e prospettive della tecnologia PLC” Rivista AEI Ottobre-Novembre 2003 [Bar&Vit2003] M.Barcaroli, N.Vitale “Telecontrollo reti BT e telegestione gruppi di misura: caratteristiche del sistema” Telecontrollo rete acqua GAS ed Elettriche 2003, Firenze, 2-3 Febbraio, 2003 [Cor et al. 2003] G.Corallini, A.Ferrara, D.Novati “Telecontrollo reti BT e telegestione gruppi di misura: Apparati di Cabine MT/BT e misuratori elettronici” Telecontrollo rete acqua GAS ed Elettriche 2003, Firenze, 2-3 Febbraio, 2003 [Cet & Cor 2003] A.Cerretti, F.Corti “Nuove soluzioni per il telecontrollo, la supervisione e l’automazione della rete elettrica AT e MT” Telecontrollo rete acqua GAS ed Elettriche 2003, Firenze, 2-3 Febbraio, 2003 [Mat&Cic 2003] M.Mattioli, L.Cicognani “Monitoraggio web delle componenti IP di un sistema di telecontrollo” Telecontrollo rete acqua GAS ed Elettriche 2003, Firenze, 2-3 Febbraio, 2003

3/3 [Mog1989] I.Moghram, S.Rahman (1989) “Analysis and Evaluation of Five Short-Term Load Forecasting Techniques”. IEEE Trans. on Power Systems, 4, 1484-1491. [Cav1998] D.W.Caves et al. (1988) “Load Impact of Interruptible and Curtailable Rate Programs: Evidence from Ten Utilities”. IEEE Trans. on Power Systems, 3, 1757-1763. [Pel1996] E.Pelikàn et al. (1996) “One-day Prediction of Electricity Load Reflecting Future RCS Schedule”. Journal of Forecasting, 15, 427-435. [Cia et al. 2004] U. Ciarniello, E. Curcio, R. Vanzan “La domanda di energia elettrica in Italia al 2030”, CESI Report being issued. Ricerca di Sistema Seconda Fase, Dic. 2004 [Cap2004a] A. Capozza “Progetto ECORET – Effetti del controllo dei carichi sullo sviluppo delle reti di BT e MT. Rapporto conclusivo del Workpackage 1 (PRECA – Previsione dell’evoluzione dei carichi dovuti ad usi finali in campo civile)”, CESI-A4510026, Ricerca di Sistema Seconda Fase, Dic. 2004 [Cap2004b] A. Capozza “Progetto ECORET – Effetti del controllo dei carichi sullo sviluppo delle reti di BT e MT. Rapporto conclusivo del Workpackage 2 (CONCA – Identificazione delle possibili azioni in grado di contrastare in modo efficace lo sviluppo dei carichi elettrici)”, CESI-A4510027, Ricerca di Sistema Seconda Fase, Dic. 2004 [Ram1995] B. Ram (1995) “Tariffs and Load Management: a Post Privatisation Study of the U.K. Electricity Supply Industry”. IEEE Trans. on Power Systems, 10, 1111-1117. Main Pictures: Figure 1: System architecture and information flow