Optimized damper control o pressure and airlow in ventilation systems Chrysanthi Soia Koulani, M.Sc. 1 Christian Anker Hviid, Assistant Proessor 1 Søren erkildsen, Ph.D. 1 1 echnical University o Denmark (DU), Department o Civil Engineering, Denmark KEYWORDS: ventilation system, HVAC control strategy, variable air volume, static pressure reset, modelling, Simulink, energy savings. SUMMARY: Conventional control strategies in variable air volume (VAV) ventilation systems do not take ully into advantage the potential energy savings since the system operation is based on maintaining a constant static pressure (CSP) set point in the main duct irrespective o the actual pressure demand. he static pressure reset (SPR) control strategy can optimize the operation o the supply air ans by adjusting the pressure set point to be just enough to deliver the reuired airlow to the most critical zone. his study investigated the operation and energy savings potential o an SPR control algorithm by using the Simulink programming tool which is add on sotware to MALAB mathematical programming language. A model o a VAV ventilation system was created in Simulink based on the InternationalBuildingPhysicsoolbox(IBP);theIBP thermal zone was remodelled in order to calculate dynamically the airlow demand according to the zone air temperature. he perormance o the Simulink model was evaluated based on the experimental setup o the ventilation system. he SPR control method established stable system operation and was proven eicient to maintain comortable space conditions while reducing by 14 % the an energy used in a typical working day. 1. Introduction In traditional control o variable air volume (VAV) systems the terminal boxes and the air handling unit (AHU) are operated independently without integration. he common practice is to control the AHU to a constant static pressure (CSP) set point corresponding to the pressure rise reuired under the design ull load condition (Wei et al 004).However in this way the AHU is regulated irrespective o the actual pressure demand. his is because under part load condition the an is providing excessive static pressure (Wei et al 004;Federspieletal005;Liuetal1997) which is dissipated by increasing the airlow resistance o the air distribution network via throttling at the terminal boxes. As a result signiicant an power is wasted in mechanical energy losses. By integrating the control o terminal boxes into the building management system (BMS) it is possible to implement the static pressure reset (SPR) control strategy. his method regulates the AHU in real time according to eedback rom several individual zones. In this case the an generates the pressure reuired in order to satisy the space conditions in the most critical zone while maintaining the airlow resistance o the distribution network at a minimum (Wang et al 1998). Conseuently the an pressure rise and thus the an power is reduced. he control method o trim and respond based on zone pressure reuest alarms is the most eicient SPR strategy since it is more stable, lexible and it minimizes the impact o rogue zones (aylor 007). he objective o this paper is to investigate the operation and the potential energy savings o the trim and respond SPR control strategy. A mathematical model o a conventional VAV ventilation system is developed in Simulink (Simulink 013) and the model is expanded by implementing the optimized SPR control algorithm. he Simulink model operation is validated by experiments perormed on the ull scale test system.
. he Simulink model he model o the VAV ventilation system was created in the graphical environment o Simulink in Matlab (Matlab 013) and it was built based on the blocks o the international building physics toolbox (IBP). he IBP toolbox (IBP 01) is a library o blocks added on Simulink, specially constructed or the thermal analysis in building physics. he construction blocks (external and internal suraces, windows) provide detailed calculations o the thermal state o every subcomponent in the structure according to the surrounding conditions to which it is exposed. he thermal condition o the zone is calculated according to the heat gained through the building envelope, the systems used or heating, ventilation and air conditioning, the internal gains occurring in the zone and the weather data corresponding to a certain location. he deault IBP blocks or the internal gains and the ventilation system were rebuilt in order to it the operation needs o the VAV ventilation system. FIG 1 illustrates the Simulink model which consists o three IBP thermal zones. Fan Friction losses Single losses hermal zone Damper Weather data FIG 1. he Simulink model o the VAV ventilation system..1 he internal gains he IBP internal gains block was conigured to include an hourly load schedule o the ventilated zone. he modiied block calculates the heat gains based on user deined proiles considering occupants, euipment and lighting use in the zone.. he ventilation system he IBP ventilation system block was remodelled to calculate dynamically the airlow demand ( dem ) according to the zone air temperature ( a );thestrategyis implemented with the ramp unctions shown in FIG. he user deined data are the minimum ( set,min ) and maximum ( set,max ) airlow set point reuired in order to maintain a comortable temperature range ( set,min, set,max ) in the zone.
dem a dem set,min set,min ( a ) FIG. he dynamic calculation o the airlow demand according to the zone air temperature..3 he an he an operation is regulated according to the tracking error determined as dierence between the CSP set point and the duct presureatthesensorposition;theblockdiagram is shown in FIG 3. FIG 3. he an operation principle. he an block receives the signal generated by the controller which corresponds to the an speed. his control signal is used to calculate the new plant output, the an pressure rise. he an characteristics (F) are represented by using the irst order linear time invariant (LI) system (Franklin et al 1993) given in euation 1. F ( s ) k s 1 Where s representation o Laplace transormation ( ) k the process gain correlating the plant input with the plant output (Pa/rpm) the time constant is the time reuired to reach the system a steady state condition (sec) he controller operation complies with euation ;the tuning o the proportional integral (PI) controller is perormed by using the simple analytic rules proposed by S. Skogestad (00). (1) n ( t ) K p, e ( t ) K i, e ( t ) dt () Where e the tracking error between the CSP set point and the pressure sensor reading (Pa) K p, the proportional controller gain, 7.14 K i, the integral controller gain, 17.85.4 he damper he damper operates as shown in FIG 4;thedam perpositionis adjusted based on the tracking error determined as dierence between the zone airlow demand (see FIG ) and the airlow provided to the zone. In the control process in FIG 4 the plant block is representing the damper system that receives the controller signal which corresponds to the damper position. he plant output, the resistance coeicient, is calculated accordingly. he second order LI system (Franklin et al 1993) presented in euation 3 approximates the operation o a typical damper (D).
FIG 4. he damper operation principle. D d n ( s ) (3) s k n s n Where k d the process gain correlating the plant input with the plant output (m 3 /s/pa%) n the natural reuency relevant to the speed response o the system, assumed 10 rad/s the damping ratio relevant to the oscillation mode o the system, assumed 1 he controller operation is applied according to euation 4;the tuning gains o the PI controller are set based on typical product values. dp( t ) K p, d e ( t ) d K i, d e ( t ) dt d (4) Where e d the tracking error between the demanded and the delivered airlow in the zone (m 3 /s) K p, the proportional controller gain, 1 K i,, the integral controller gain, 10.5 he pressure and airlow solver he riction and single pressure losses blocks illustrated in FIG 1 implement the hydraulic calculation o the VAV ventilation system according to the duct design shown in FIG 5. he unknown pressure and airlow conditions are determined by setting up a system o euations expressing the pressure losses occurring in every component o the system. he hydraulic calculation determines the pressure demand (P) at the beginning and end o every component as well as the airlows () delivered to the dierent zones (see FIG 5). he system o euations cannot be solved analytically;thereorethe Newton Raphson numerical method is used instead. FIG 5. he duct design in the pressure and airlow solver block. In a piece o duct the pressure losses due to riction are calculated according to the Darcy Weisbach euation (White 1998) given in euation 5. he Darcy riction actor is obtained by the Swamme Jain euation (Swamme et al 1976), which is an approximation o the implicit Colebrook White euation (see euation 6).
P r D L d u air Where D the Darcy actor ( ) L the length o the duct (m) d the diameter o the duct (m) D log 10 0.5 5.74 0. 9 Re 3. 7 d Where Re the Reynolds number ( ) the roughness height, or thin plate ducts is eual to 0.15 10 3 m he pressure losses due to connections and ittings are calculated according to euation 7. (5) (6) P sing r u air Where r the resistance coeicient ( ) u air the mean velocity o airlow (m/s) the density o air, 1.04 kg/m 3 he pressure losses introduced by the damper component are approximated based on euation 8. (7) P sing k del value Where del the airlow delivered to the zone (m 3 /s) k value the damper resistance coeicient (m 3 /spa) (8).6 he static pressure reset algorithm In order to implement the SPR control method o trim and respond based on zone pressure reuest alarms, one more block was added to the Simulink model presented in FIG 1. he operation principle o the applied control logic can be seen in FIG 6. FIG 6. he control logic o the trim and respond static pressure reset method. Every damper o the VAV system transmits an alarm signal when its position exceeds 85% opening; the zone keeps generating a pressure reuest until the damper closes to a position o 80 % opening. he pressure reuests rom all zones are summed and when at least two zones give an alarm the an pressure set point is reset 10 % upwards o the pressure demand at the sensor position. In the opposite case it is reset 5 % downward. he SPR is perormedwithinaspeciicpresurerange;theupperlimit is set eual to the CSP set point while the lower SPR limit is determined according to the pressure demand ensuring precise damper operation. he SPR loop resets the pressure set point every 90 sec
and the an adjusts to the new pressure set point. 3. he experimental setup he perormance o the Simulink model was validated on a ull scaleexperimentalsetup;the experimental setup arrangement was identical to the duct design given in FIG 5 where the distribution duct had a diameter o 315 mm and the connection ducts a diameter o 160 mm. he setup consisted o three LeanVent DropDamper LERX and a box an (Exhausto BESF1804 1EC). he VAV system was evaluated both with the CSP and the SPR control strategy;the two methods were modelled in LabVIEW (LabVIEW 013). he validation was perormed by providing the Simulink model and the experimental setup with the same airlow demand data;two dierent airlow demand proiles were used or testing the model perormance with each control strategy (see FIG 7). Ventilation zone 3 behaved like a rogue zone with the SPR method because the lower limit o the pressure range, within which the an operation set point reset, was insuicient or delivering the reuired pressure. CSP demand proile SPR demand proile FIG 7. Airlow demand proiles with the CSP and the SPR control method, respectively. 4. Results 4.1 Simulink model validation he graphs in FIG 8 compare the perormance o the Simulink model and the experimental setup o the VAV ventilation system when the an was controlled with the CSP and the SPR control method, respectively. FIG 8. Response curves o the dampers with the CSP and the SPR control method, respectively.
According to the obtained results the Simulink model with both control strategies approximated satisactory the actual operation conditions since the dampers ollowed a similar response trend. he largest deviation between the Simulink model and the experimental setup was obtained rom damper when the VAV system was regulated with the SPR control method. As presented in the second graph in FIG 8 the deviation was below 10 %, however in the last minutes it increased to %. his occurred because the damper modelling represented insuiciently the actual speed response o the Dropdamper. he mathematical model o the damper should be tuned to coincide with the experimental data. he speed o the damper model is relative to the natural reuency parameter involved in the mathematical expression describing its response (see euation 3). In this case the speed o the damper wasasumed; in order to accelerate the response, trial and error simulations have to be perormed increasing the natural reuency. 4. Energy savings rom optimized damper control he energy savings potential o the SPR control strategy is determined based on the an power used when controlling the ventilation system with the CSP and the SPR method, respectively. Considering that the an eiciency varies according to the dierent pressure and airlow conditions that the an is operating with, the an eiciency was approximated rom producer table values by using average hourly values o airlow and pressure rise delivered by the an. he average hourly values were derived based on 4 hour simulation data obtained rom the Simulink model o the VAV ventilation system when operated with both control strategies. he an power was calculated according to euation 9. Where Power P an n e tot Pan the an pressure rise (Pa) tot the an airlow (m 3 /s) n e the an eiciency ( ) he an energy used in a typical working day in winter when the occupancy period was set rom 8 am to 17 pm with the CSP and the SPR was determined to 151 Wh and 130 Wh, respectively;theenergy consumption was reduced approximately by 14 %. he highest energy savings potential o the SPR control method occurred under part airlow conditions as the an operated with decreased static pressure. Due to the act that the an was correctly sized the combination o lower pressure set point and airlow improved an eiciency and thus the energy savings were urther increased. With the CSP control method the same airlow conditions combined with increased static pressure lowered the an eiciency and as a result the an operated ineiciently. 5. Conclusions he results presented in this paper draw the ollowing conclusions: he irst and the second order LI systems were proven representative or the response o the an and the damper, respectively. For the current study we assumed the parameters o the mathematical model o the damper (, n). he value o the natural reuency parameter turned to be inaccurate or representing the speed response o the Dropdamper as it introduced high deviation to the Simulink model when it was controlled with the SPR control. he damper mathematical model should be tuned to it the experimental data. he SPR control method documented higher energy savings under part airlow conditions where the an operated with decreased pressure set point. In practical applications o the SPR control strategy caution should be given when determining the lower limit o the pressure range within which the variable an operation set point is established. his (9)
parameter is critical because in case that the minimum an pressure rise is insuicient to satisy the system pressure demand, the ar located zones will act as rogue zones. hereore it is advisable to perorm airlow measurements in order to ensure that the set point selected is appropriate or the precise operation o the dampers. In order to maximize the energy savings potential o the SPR control method, the an should be correctly sized;inthiscasetheaneiciencyimproveswhen the an operates with decreased static pressure. 6. Acknowledgements he authors acknowledge the inancial support rom the Energy echnology Development and Demonstration Programme (EUDP), Danish Energy Agency. Moreover the authors wish to express their gratitude to Remus Mihail Prunescu, Christos Papoutsellis and Vasilis Bellos. Reerences Federspiel CC, Haves P. & Cohen. 005. Detecting critical supply duct pressure. ASHRAE ransactions, 111 PAR 1, pp957 63. Franklin G.F.;Powell J.D. & Naeini A.E. 1993. Feedback control o Dynamic Systems. nd ed. Addison Wesley Longman Publishing Co., Inc. Boston, MA, USA. IBP. 01. International Building Physics in Simulink. Available rom: http://www.ibpt.org/ [accessed 01 0 013] LabVIEW. 013. National instruments, version 010. Available rom: http://sine.ni.com/np/app/main/p/docid/nav 104/lang/da/ [accessed 05 08 013] LiuM.;Zhu Y.;ClaridgeD.E.& W hitee.1997.impacts o Static Pressure Set Level on HVAC Energy Consumption and Indoor Conditions. ASHRAE ransactions, 103, (). Matlab. 013. Mathworks, version 01b. Available rom: http://www.mathworks.com/products/matlab/ [accessed 14 10 013] Simulink. 013. Mathworks, version 01b. Available rom: http://www.mathworks.com/products/simulink/ [accessed 14 10 013] Skogestad S. 00. Simple analytic rules or model reduction and PID controller tuning. Journal o Process Control, 14, (4), pp 465. Swamee P.K. & Jain A.K. 1976. Explicit euations or pipe low problems. Journal o the Hydraulics Division (ASCE), 10, (5), 657 664. aylor S.. 007. Increasing Eiciency with VAV System Static Pressure Set point Reset. ASHRAE Journal 49, (6), pp4 3. Wang S. & Burnett J. 1998. Variable Air Volume Air Conditioning Systems: Optimal reset o static pressure set point. Building Services Engineering Research & echnology, 19, (4), pp19 31. Wei G.;LiuM. & Claridge D.E. 004. Integrated damper and pressure reset or VAV supply air an control. ASHRAE ransactions, 110, (). White F.M. 1998. Fluid Mechanics. 4th ed. McGraw Hill Editions, 79 pp.