12th International Symposium on District Heating and Cooling

owner is normally only interested in lowering theenergy consumption, while the district heating companyis more interested in being able to optimize the wholeproduction and distribution process. Optimizing theproduction normally translates to avoiding expensiveand, more often than not, environmentally unsoundpeak load boilers or trying to move heat load demandin time in order to maximize utility during combinedheat and power generation. Basically, from theperspective of the district heating company it is aquestion of finding a balance between loweringexpensive heat load demand while still selling as muchenergy as possible. Implementing this on a systemwide scale requires complex coordination controlstrategies that dynamically adapt to the state of thedistrict heating system [2]. On the local building levelthis is implemented by performing temporary heat loadreductions. On a local level these reductions arenormally very short, i.e. one or a few hours, but theycan be of high intensity, even sometimes completelyshutting of the heat load during shorter periods of time.This behaviour requires the control system to be highlyadaptive in relation to the dynamics of the buildingsthermal inertia in order to avoid jeopardizing the indoorclimate. By coordinating such local heat loadreductions among a large group of buildings it ispossible to achieve system wide DMS and LC.Previous workMost previous work regarding temporary heat loadreductions deals with night time set-back. This is atechnique that has been around for a long time, and isbased on the general idea that if you decrease thedifference between the outdoor and indoor temperaturein a building you will save energy. One of the firstlarge-scale evaluations of night time set-back wasperformed in 1983 when buildings in Sweden, USA,Belgium and Denmark were evaluated. Thisexperiment concluded that night time set-back did notsave as much energy as was expected, at most a fewpercent for multi-apartment buildings [3]. In hindsight itis possible to see that these meagre results were aconsequence of several interacting factors. First of allthe control systems of the time were not capable ofproperly handling the transition from night time setbackto the original operation mode, which causes aconsiderable over-compensation of heat load when thesystems tries to find the new control level. This extraboost in heat load during the mornings counteractslarge portions of the energy saving done during thenight. The theoretical part of the experiment also had afew draw-backs, e.g. assuming optimally adjustedradiator systems and linear relations between indoortemperature and energy savings. Other articles showthat there is indeed a substantial level of energy savingto be found by controlling the local heat load [5].The <strong>12th</strong> <strong>International</strong> <strong>Symposium</strong> on District Heating and Cooling,September 5 th to September 7 th , 2010, Tallinn, Estonia245Most of the previous work done on the subject is basedon simulated results. This is expected since thedynamic thermal processes within a building areextremely complex and it is not surprising thatcomparisons between measurements and calculationssometimes show large discrepancies. It is noted thatmost calculations are dependent on variables thatcannot be measured and verified, and that the buildingtime constant is really not a constant [6].EXPERIMENTAL METHODIn order to study the effects of temporary heat loadreductions we equipped a building with several wirelesstemperature sensors in order to measure thefluctuations in indoor temperature. The building inquestions is an office building with semi-light thermalcharacteristics (light construct with concrete slab) and atime constant of about 150 hours [7]. The indoortemperature sensors were placed on different locationswithin the building in order to get a good overview ofthe thermal behaviour of the indoor climate. In additionto the existing outdoor temperature sensor an extrawireless sensor was also placed on the outside of thebuilding. Unlike the existing outdoor temperaturesensor the wireless one was placed in a position wereit was fully exposed to any possible sunshine. Thisgave us an extra indication of the impact of free heatingthrough window areas, even though we did not haveany ability to measure the actual solar irradiance.In order to control the district heating consumer stationwe connected a load control platform for system wideLC and DSM [8]. This platform is based on a novelform of hardware and software which enables us tomanage the heat load of the substation without anymajor alterations or any damage on the existinghardware. The software system is based on the opensource Linux operating system and is equipped with anapplication programming interface (API) for I/O. Thismakes it easy to apply additional sensors, e.g. formeasuring the forward and return temperatures of theradiator system. The platform also featuresconnections to a database system which enables realtimelogging and analyse of sensor data. The actualheat load reductions are implemented by supplying theexisting control system with adjusted outdoortemperatures, which gives us the ability to manage thebehaviour of the heat load without exchanging anyexisting hardware. This adjusted outdoor temperaturecan be managed with a resolution of at most 60seconds. The computer platform uses either Ethernetor GPRS modems to communicate with the database.In our case we used the existing Internet access in thebuilding. In addition to this primary experimentalbuilding we also collected and analysed data frompreviously installed buildings using the same basiccomputer platform.