The TOU unfrequently schedule and consider renewable sources. updated does not Ripple Control Based on specific signal frequency over the power broadcast lines. at Acknowledgements Remote control of larger group of devices only. The most common management method direct used in the load since 1980s. This work industrial has cooperation been Mycroft Mind company with as a part of done the in the Lasaris smart activities. grid reasearch

Algorithm for calculation of the calculation Algorithm for water heating TOU tables for appliances. Load Controller Component Architecture Appliance switching is scheduled one day in advance. Integration of local weather conditions. Customer's participation is rewarded with lower tariffs. Load ManagementLoad Practices Load management aims at flattening the load curve by either increasing or decreasing the load. Our approach allows individual control of the water heating appliances. Local Load Control

Achieved by and indirectappliance control) methods (Time‐Of‐Use tariffs). combination of direct methods (remote

Hradec Králové ~500 Vrchlabí ~5 000 Representation of the transmission of the transmission Representation metric in the ideal case (red), losses the local load algorithm(green) case and actually observed case (blue). Example of daily TOU schedule. Rows represent Rows represent of daily TOU schedule. Example individual water heaters while columns represent the Green squares show of time. periods 15 min. which the water heaters are during periods on. switched Jeřmanice Pardubice ~5 000 ~25 000 Our Load Controller component has been deployed in the production environment of the major company in the Czech Republic since March 2015. power distribution Smart Grids in Grids Smart theRepublic Czech Three major distributors. The estimated reduction of transmission losses is average (7.04% maximum, 2.28% minimum). 4% on It is used in three consumption points with water heating appliances. urban areas, each consisting of ~100 Results Case Study The algorithm is executed daily and it constructs the TOU tables for water heaters control which are then uploaded to the smart meters. Both industrial and residential customers. Local pilot projectsLocal pilot of deployments 2010. since Participation in the international Grid4EU project. Total of ~38 000 smart installations. of ~38 000 Total The distribution of losses is between 95% and 97% optimal distribution. of the

[2] Fenton, N., Neil, M. Risk Assessment and decision analysis with Bayesian networks. Crc Press, 2012. [3] Neuberg, A. Ripple control in the Czech Republic and demand side management, CIRED, 2009. [1] Kostková, K., et al. An Introduction to Load Management. System Research 95 (2013): 184‐191. References

Comparison of old from from of oldComparison load profile the ripple control and the (black) the from local load profile predicted load optimization algorithm (green).

The nodes are initialized with historical consumption data collected from smart meters. We We use them to predict water heating appliances. regulated consumption of the Integration of weather forecast. There is total of 96 Bayesian networks points, each per representing consumption a 15 min. period of single water heater. a day for a Allows incorporation of uncertain data and propagation of observed evidence. Challenges Power distribution companies want with minimalpower load transmission losses. to ensure stable Current load management practices in the Czech Republic are flexible not enough to adapt in the rapidly changing environment. Probabilistic graphical model. Compliance with rules issued by Energy Regulation Office. Lack of reliable data. Growing number of customers utilise renewable power sources via solar or plants which makes them contributors to the grid as well. Bayesian Networks Goals Smart Grid technologies enable collection of useful data that can be used to predict future consumer behaviour. Usage of data power about consumer consumption from smart meters. collected Motivation for Smart Management Load Implementation of software component and algorithm for localised load transmission losses. management that would minimize