Computer Architecture for Embedded Systems (CAES) group Faculty of Electrical Engineering, Mathematics and Computer Science University of Twente Enschede, The Netherlands November 13, 2008 Controlling the energy production at home Maurice Bosman PhD TW colloquium
November 13, The year is AD 2008… And electricity is entirely produced by large power plants. Well not entirely! There is a strong trend towards a distributed elec- tricity production.
November 13, Energy market Liberalised on July 1, 2004 Competition! Electricity producers vs electricity suppliers meet on electricity market (APX) Grid operators are obliged to allow all suppliers in their region Entrance possibility: distributed production
November 13, Distributed electricity production Less transport losses Higher efficiency of production at home Use of renewable sources CO 2 reduction Relief of loads of electricity grid Production capacity limited Demand/supply matching
November 13, Electricity production at home
November 13, Heat production at home
November 13, MicroCHP Micro Combined Heat and Power Input: gas Output: electricity and heat Electricity consumed at home or delivered to the grid Heat consumed at home (no concrete plans to share heat within a neighbourhood)
November 13, Heat demand in a house Central heating, tap water Immediate supply by household device (unless you live in Enschede Zuid) Heat buffer necessary for scheduling
November 13, Electricity demand in a house Fridge, tv, coffee machine, … Supply is no issue (unless you live in Haaksbergen) Electricity pricing Electricity buffer possible, but not necessary
November 13, radiator If you live in Haaksbergen grid microCHP 10
November 13, Problem setting Use a microCHP in house Apply this onto many houses Electricity supplier offers global control of the appliances
November 13, Research goal Study the consequences of introducing a fleet of microCHPs: Controllability/scalability Optimization heuristics
November 13, Controllability/scalability Global Scheduler Local Scheduler (Embedded Computer) Hierarchical structure Hard Constraints: Household comfort Limited communication Real time decision making
November 13, Optimization heuristics Several objectives Minimize total electricity costs Minimize total energy costs Maximize total electricity revenue Minimize transportational losses Minimize peak loads at transformers Make use of electricity market (APX) Make use of electricity and heat profiles
November 13, APX prices
November 13, Scheduling Offline: use all available information Online: receive a job and schedule immediately (example: earliest possible)
November 13, Our scheduling problem Jobs : switched on microCHP appliances Jobs have undetermined length! Online problem; repetitive jobs
November 13, ILP formulation Decision variable ‘Accountancy’ equations
November 13, ILP formulation Objective: minimize/maximize something Heatstore; below LL: switch on Heatstore; above UL: switch off
November 13, ILP formulation Need to run minimum runtime MR Stay switched off for minimum time MO Fleet capacity restrictions
November 13, Scheduling problem Optimal values (AIMMS)
November 13, Scheduling problem When is it good to use longer jobs? Divide jobs into classes Heatstore information Runtime information Consumption prediction Make decisions that balance classes! Make switch off decisions!
November 13, Questions ?