12th International Symposium on District Heating and Cooling

The <strong>12th</strong> <strong>International</strong> <strong>Symposium</strong> on District Heating and Cooling,September 5 th to September 7 th , 2010, Tallinn, EstoniaYes11:50 PM < time < 0:00 AMWait 60 min withconstant pump speedStart timerWait 5 min< 80 min?NoSave currentvaluesSet controlvalve to AutoNo,> 80 minGet T o , usemodified controlcurveSet pump speedfor testSet TPump speed 1?YesWait 10 minTest pumpspeed 1Wait 10 minTest pumpspeed 2Set controlvalve to Manualpoint, one could expect a stable secondary returntemperature, e.g., during the last five minutes. Inaddition to the secondary supply temperature, also theprimary supply temperature is recorded. However, thedampened outdoor temperature, i.e., the input signal tothe controller, is recorded when the CV is locked for thefirst time. The reason for this is that the heat supply issubsequently kept constant at a level matching theoutdoor temperature (and heat load) at the time beforethe test was started.NoYes No, pump speed 2T – T set-p > 0.2°C Yes T set-p = T s,r + T set-pReject testresultUpdatecurvesNot okOkCheck maximumdeviation for Q(e.g., 5%) and To(e.g., 2°C)Test doneDetermine T p,r,rad,min(pump(0), pump(1)or pump(2))Fig. 4 Flow chart describing the adaptive control algorithm.If a modified control curve is used before a test is aboutto start, the control should be interrupted and the pumpspeed kept constant for an hour prior to the test. Thisway, one avoids the risk of the flow changing (due toalterations in the outdoor temperature) too close to thetest, which could result in unstable radiator systemtemperatures.The supply and return temperatures were measured onfour of the most remote risers from the substation,during the tests. A continuous matching againstmeasurements on risers gives a good indication thatthe flow distribution in the system was not impaired bythe optimisation. The temperature profile was closelymatched to the profile at the substation. Both flowreductions resulted in increased temperature drops.Updating the control curvesAfter the completion of a test, the obtained informationneeds to be evaluated. The influence of the variation ofthe outdoor temperature is not entirely obvious; itsinfluence decreases with an increasing time constantfor the building. Variations on the primary side normallyhave is compensated for since the heat supply is keptconstant. As a result, it is sufficient to verify that theheat supply was maintained at a steady level during thetest, avoiding any disruptions.If a test result is accepted, the primary returntemperatures for each tested flow are compared inorder to verify which flow resulted in the lowest returntemperature. This flow also gave rise to a secondarysupply temperature. It is however not obvious how toread this temperature, given that it was regulated bythe controller and changed continuously. The mostlogical choice is to read the mean value at the end ofthe test period, before the pump speed changes. At thisThe next step consists in using the information attainedfrom the test to modify the control curves. Initially, theoriginal curve was used and the pump was, in ourcase, controlled to give a constant differential pressure.If the result of a test is that a lower primary returntemperature is obtained at a lower secondary flow rate,the control curve is updated for that outdoortemperature. A reasonable resolution is 1 °C. Theoriginal control curve, generally based on 5–8 points,was therefore initially extended to comprise values foreach outdoor temperature.If the experiment, as in Fig. 3 above, was performed at8 °C, this point on the curve would be updated. Alongwith the new supply temperature there followed a newradiator flow, which in our case was expressed as anew set-point for the pump speed.The adaptive control continues in this manner nightafter night, and the control curves are continuouslyupdated. Outside the test periods of approximatelythree hours each night, the modified control curves areused for controlling the heating system.Fig. 5 shows an example of the gradual development ofthe modified control curve. The first graph shows a newpoint at 0 °C (used for 0 ± 0.5 °C). In the second(upper) graph, a point for 3 °C has been added, whilethe range 0 to 3 °C is complete in the third. The fourthgraph shows a much more complete control curve(-5 to 10 °C). Temperature curves corresponding toconstant flow systems with lower flows than the originalsystem have been included as thinner lines. The valuefor 10 °C coincides with the curves of a system with alow flow, while the value of -5 °C coincides with thecurves of a system with a moderately reduced flow(normal flow). The last graph clearly demonstrates thatthe modified curves are based on a variable flow, i.e.,they coincide with various constant flow curves atdifferent points.As shown in the second graph of Fig. 5, the modifiedcurve could emerge in sections that subsequently arecombined. One way to speed up the modification of thecontrol curves is to interpolate intermediate valuesrather than wait for a flow optimisation at the missingoutdoor temperature. Even the return temperaturescould be interpolated, since it is possible to determine211