DEVELOPMENT OF A FAN AIRFLOW STATION FOR AIRFLOW CONTROL IN VAV SYSTEMS

Size: px

Start display at page:

Download "DEVELOPMENT OF A FAN AIRFLOW STATION FOR AIRFLOW CONTROL IN VAV SYSTEMS"

Easter Douglas
5 years ago
Views:

CONTROL IN VAV SYSTEMS Dvid Yuill, University of Nebrsk-Lincoln

1 University of Nebrsk - Lincoln From the SelectedWorks of Dvid Yuill Spring Mrch, DEVELOPMENT OF A FAN AIRFLOW STATION FOR AIRFLOW CONTROL IN VAV SYSTEMS Dvid Yuill, University of Nebrsk-Lincoln Nthn Redmnn Mingsheng Liu Avilble t:

2 Proceedings of ISEC Interntionl Solr Energy Conference Hwii, USA, 5-8 Mrch ISEC-44 DEVELOPMENT OF A FAN AIRFLOW STATION FOR AIRFLOW CONTROL IN VAV SYSTEMS Dvid P. Yuill, Nthn K. Redmnn nd Mingsheng Liu Ph.D., P.E. Energy Systems Lbortory Architecturl Engineering University of Nebrsk South 67 th Street Omh, NE 688 United Sttes of Americ ABSTRACT A fn irflow sttion hs been developed by the uthors. This sttion is used to determine fn irflow using fn speed nd fn hed s inputs. These inputs cn be mesured ccurtely in vrible ir volume (VAV) systems, so they cn be used s cost-effective control monitoring method for heting, ventiltion nd ir conditioning (HVAC) systems. Theoreticl models hve been developed for the flow sttion using both second nd third order reltionships between the irflow nd the inputs. The theoreticl model hs been experimentlly tested nd excellent greement between the model nd the experimentl vlues ws found. The second order model ws within.5% of the experimentl vlues, nd the third order model ws within.7%. The third order model is much more complex thn the second order model, nd offers no dvntges. INTRODUCTION With the energy crisis of the 97 s push ws mde by engineers to design more energy efficient HVAC systems. It ws during this time tht vrible ir volume (VAV) systems begn to be used commonly, rther thn constnt ir volume (CAV). By modulting the supply irflow these systems supply less cold ir to rooms which do not need s much cooling, thus sving rehet energy. With severl terminl boxes modulting to stisfy cooling lods, the overll systems hd vrying flow requirement. There were severl wys to ccommodte this requirement. Simply dding flow resistnce in the ductwork would reduce the flow. However by vrying the speed of the fn, or by deploying inlet guide vnes, fn energy could lso be sved. The more efficient of these solutions is to vry the fn speed [, ]. This is typiclly crried out using vrible frequency drive (VFD), device which modultes the frequency of the lternting current electricl signl. However, CAV systems, though less energy efficient, hve n dvntge in tht they re esily controlled. After the initil testing nd blncing the building cn hve the proper outside irflow rte estblished, nd the proper difference in the supply nd return irflow rtes, thus ensuring proper building pressuriztion. Building pressuriztion is required to eliminte infiltrtion []. Infiltrtion is uncontrolled irflow into building, nd it cn hve mny negtive effects on the building. The most importnt is tht infiltrtion brings moisture into the building during the cooling seson. This moisture cn condense, cusing dmge to the building structure, to insultion, nd promoting growth of mold nd mildew [4]. Other detriments of infiltrtion cn include locl therml comfort problems (in exterior zones, prticulrly during very cold wether) nd introduction of ir pollutnts into the building. Conversely, if ventiltion ir enters the building in controlled fshion it cn be drwn from n re isolted from locl sources of pollution, then filtered nd dehumidified before it enters the occupied spce of the building. To ensure tht there is no infiltrtion, the mechnicl system should supply more ir to the building thn it returns nd exhusts. This difference between supply flow rte nd combined exhust nd return flow rtes is the exfiltrtion flow rte. Q = Q Q + Q ) () xf s ( xh r The exhust mount is normlly constnt, nd the desired exfiltrtion flow rte is constnt. The sitution becomes complicted when the supply nd return mounts vry. Three building pressuriztion control methods hve been developed over the lst yers. These methods re fn trcking, direct building pressure control, nd volumetric trcking. The fn-trcking method (sometimes referred to s VFD proportionl drive-slving) involves link between the

3 VFDs controlling the supply nd return fns. The supply fn is normlly controlled to keep constnt sttic pressure t some remote point in the ductwork. When less supply ir is needed the supply fn slows down nd the return fn lso slows down. The problem is tht the system flow resistnce chrcteristics chnge on the supply side (the terminl box dmpers), but do not chnge on the return side. This mens tht the supply fn is running t different speed nd is encountering different pressure thn t design conditions, so tht flow is not proportionl to speed. The return fn, however, is operting t fixed flow resistnce, so tht link between the speeds of the return fn nd supply fn will not mintin correct building pressure except t the point for which the system is blnced [5-7]. The direct building pressure control method mesures the building pressure with reference to outdoor pressure, nd uses this s control input for the return fn. If the building is under-pressurized the return fn slows down so tht less return ir is drwn bck from the spce, nd less ir is sent to the relief ir duct. In theory this is n excellent solution, but in prctice the problem comes in the mesurement of building pressure. It is very difficult to ccurtely mesure such smll sttic pressure differences s re required for building pressuriztion (typiclly.5 in. w.g. ( P) to. in. w.g. (5 P) depending on climte [8]). The sensors need to be clibrted, nd necdotl evidence suggests tht very few opertions mngers know the loction of the pressure sensor. If the pressure sensor is subject to wind then it won t serve its purpose. Also, the pressure my vry throughout the building, prticulrly in multi-story building. This method, using existing technology, is extremely difficult to implement. Volumetric trcking is control method tht cn be effective under certin circumstnces [6, 9-]. It involves putting flow sttion in the min supply duct nd nother in the min return duct. The return fn speed is controlled by compring the flow rtes in the supply nd return ducts. The fn is modulted to keep this difference t constnt setpoint, ensuring constnt exfiltrtion rte (ssuming exhust s constnt). The problem with this method lies in the flow sttion mesurements. For ccurcy within 5 to % flow sttion normlly needs stright, unimpeded duct run for 6- dimeters upstrem nd severl downstrem of it []. There re very few systems tht hve such duct runs in the min supply nd return ducts. Therefore, to properly control building pressuriztion fn irflow sttion tht cn ccurtely determine irflow cross fn provides vible solution. This pper presents n experimentl verifiction of fn irflow sttion model proposed by Liu []. The fn irflow sttion uses fn speed nd fn hed s inputs, with secondorder reltionship to fn irflow. An extension of this fn irflow sttion is lso presented in this pper, using n ssumption of third-order reltionship between fn hed, fn speed, nd flow rte. The experimentl results re compred to the second order nd third order models to ssess the vlidity of the fn irflow sttion. The fundmentl theory, experimentl equipment nd procedures, nd experimentl results re discussed in detil in this pper. THEORY The fn irflow sttion includes differentil pressure trnsducer nd fn speed trnsducer. The fn speed trnsducer my be replced by the control system s commnd to the VFD. The fn irflow sttion cn be implemented using typicl Energy Mngement nd Control Systems (EMCS). Figure presents schemtic digrm of the irflow mesurement nd control in typicl VAV system using fn irflow sttions. The supply irflow is determined using the mesured fn hed nd the VFD speed. The return irflow setpoint is then determined by tking the difference between the supply irflow nd the sum of the building exhust nd exfiltrtion flow rtes. The return irflow rte is clculted using the mesured return fn hed nd speed. The return fn speed is modulted to mintin the return irflow setpoint. The fn irflow sttion is developed bsed on the following theories. The fn hed cn be regressed s function of the irflow using the design fn curve. Eqution () is typicl polynomil regression curve. H = n d i i= Q i d where H d nd Q d re the fn hed nd irflow rtes for the fn t % speed. Relief Air Dmper DM DM Outside Air Dmper Return Air Dmper H r VFD H s VFD Nr Ns Qr Conditioned Spce Figure. Digrm of Airflow in VAV Systems using Fn Airflow Sttions When the fn is operting t reduced speed, the fn hed nd irflows cn be determined with Eq. (), bsed on the fn lws. n i Q H = ω i () i= ω where N ω =, H = ω H d, nd Q = ωqd. N d Qs () Qxh When third order polynomil regression eqution is used to represent the design fn curve, then fn irflow cn be predicted using the mesured fn hed nd the fn speed with Eq. (4). The regression coefficients re lbeled ( i ), Qxf

4 Q q + q + q q = + (4) where q = 7 4 p + ω H nd 4 p p = Using second order polynomil regression of the design fn curve, s developed by Liu [5], the fn irflow cn be predicted using the mesured fn hed nd fn speed with Eq. (5). Q = 4 H ω ω In order to provide high flow resistnce within the system reinforced sheet metl plte ws put over the entrnce to the supply duct, nd flow resistnce ws djusted by shifting this plte. For lower flow resistnce rnges the dmpers in the distribution system were used. The mesured prmeters included irflow, fn speed, sttic pressure, nd ir temperture. The irflow ws mesured using digitl flow-hood with specified precision of ± % nd ccurcy of ± 7 cfm (. m³/s). The differentil pressure ws mesured using digitl pressure sensors with precision of ± % nd ccurcy of ±.5 in. w.g. (.5 P). The fn speed ws mesured using lser tchometer (ccurcy: ±.6%; precision: ± rpm). The ir temperture ws mesured using thermometer with n ccurcy of ± F (.56 C). During ll experiments, the ir temperture ws mintined within rnge of 7 F to 74 F (. to. C). (5) The experiments conducted included: ) Fn speed/emcs clibrtion b) Genertion of design fn curve c) Fn irflow sttion verifiction EXPERIMENTAL METHOD AND SETUP The objective of the experimentl portion of this reserch ws to verify tht the theory holds true. The experiments were crried out on n ctul ir hndling unit (AHU) instlled in building systems lbortory. The AHU, shown in Fig., hs two sets of heting nd cooling coils nd bnk of filters ( low efficiency pre-filter nd high efficiency finl filter) in the (4 m) length between the flow hood nd the supply fn. Distribution System Pressure Flow Sensor Regultor P C.C H.C C.C H.C EMCS VFD Fn Filter Flow Hood Figure. Digrm of the AHU used in the Experiment The supply fn is controlled by VFD, which is controlled by n EMCS. The fn is direct drive with bckwrd-inclined bldes on.4 (. m) wheel. The mximum fn speed is 45 revolutions per minute (rpm) with brek horsepower of.75 hp (5 W). The design irflow is 479 cfm (.4 m³/s) t. in. w.g. (8 P). To mesure irflow through the system, the mixing chmber ws removed nd frme ws built round the return (suction) side of the AHU. A flow hood ws rigidly ttched nd seled to this frme to ensure consistent positioning of the flow hood. Two differentil pressure sensors were instlled to mesure fn hed (the second one for redundncy). The first of these experiments ws to determine the correltion between the EMCS commnd to the VFD nd the ctul fn speed, since the VFD speed will eventully be represented by the EMCS commnd. The EMCS commnd ws set in 5% increments from % down to 5%, then bck up to %. The fn speed ws mesured using the digitl lser tchometer. The design fn curve is defined s the correltion between fn hed nd fn irflow when the fn is running t its mximum speed. For experiment b), the fn speed ws controlled t,45 rpm. The irflow ws incresed from 5 cfm to,5 cfm (.8 to.8 m³/s) by modulting terminl box dmpers nd the flow resistnce device instlled for this study. The irflow through the fn ws mesured using the flow hood. The fn hed (the differentil pressure mesured cross the fn) ws mesured with digitl pressure sensor. For this nd ll prts of the study ech flow nd pressure mesurement ws tken over period of time, nd timeverged. Corresponding flow nd pressure mesurements were lwys mde simultneously (over the sme time period). A totl of 5 mesurements were tken in this portion of the study. For prt c) of this study verifiction of the fn irflow sttion model tests were conducted. These included fn speeds (5% to %) for ech of flow resistnce configurtions. At full speed, the flow rtes chieved with these resistnces rnged from 5 to, cfm (.54 to.4 m³/s). In ech test, the EMCS commnd, the fn hed, nd the irflow were mesured nd recorded. The fn speed ws clculted using the EMCS commnd nd the previously mentioned correltion fctor. The fn hed nd fn speed were then used s inputs to the fn irflow sttion model, nd the clculted fn irflow ws compred to the mesured fn irflow to verify the theoreticl model. Finlly, error nlysis ws conducted. The flow hood error 7 cfm (. m³/s) constitutes the mximum irflow mesurement error. The mximum fn irflow sttion error is determined to be cfm (9.4 x -4 m³/s) (See ppendix for clcultion detils).

5 RESULTS ) Fn speed/emcs clibrtion The correltion of ctul fn speed with the EMCS commnd to the VFD is plotted in Fig.. The EMCS commnd vries from 5% to % while the fn speed vries from 5 rpm to 4 rpm. At ech point on the grph there re two dt (one on top of the other): one from the incresing fn speed series, nd one from the decresing. These dt show very relible liner reltionship between the EMCS nd the ctul fn speed. There is no mesurble hysteresis on this system. The regression line gives vlue to relte EMCS speed with ctul fn speed. When the EMCS commnd is, the VFD output is 7 Hz, nd the fn speed is 6 rpm. b) Genertion of design fn curve The design fn curve tht ws generted is shown in Fig. 4. The third-order polynomil regression fits the dt very well. The dt in the higher rnge were tken by opening nd closing dmpers in the distribution system. Quite few of the 5 dt re in this rnge. In the lower flow rnges it becme difficult to crete sufficient flow resistnce becuse of duct lekge in the distribution system. For the lowest flow rtes the supply duct ws covered over with bord nd seled with putty, nd for incresing flow rtes the putty ws removed nd the bord ws shifted slightly. Fn Speed (rpm) y =.758x R = EMCS VFD Speed (%) Figure. Fn Speed s Function of EMCS Commnd c) Fn irflow sttion verifiction The gthered dt re shown in Fig. 5, below. Some best-fit lines hve been drwn through dt long lines of equl fn speed to clrify the shpe of the fn curve t different speeds. The different flow resistnce configurtions cn be seen s upwrd-sloping curved lines. One of these is doubled. This ws n experimentl check of the repetbility of the mesurements. There re flow resistnce lines. Figure 6 compres the mesured dt with the vlues predicted by the fn irflow sttion second order model. (A plot of the third order model would show no discernble difference). The plot shows the mesured irflow rte on the bsciss nd the fn irflow sttion irflow rte on the ordinte for ech pressure mesurement. The dt hve very good correltion in the upper rnges, but strnge reltionship is observed mongst mny of the dt in the low flow rnge. These nomlous dt re the result of the tests involving very high flow resistnce. This mens tht they fll within the unstble region of the fn curve. They re used to determine where the strt of the stble operting rnge occurs. Those dt which fll outside the stble operting rnge re ignored in this study. Pressure cross fn (in.w.g.) Fn Hed (in. w.g.) Airflow (m³/s) y = -8.86E-7x E-4x E+ R = 9.98E Airflow (cfm) Figure 4. Mesured Fn Curve t Full Speed Flow (m³/s) Flow (cfm) Figure 5. Mesured Airflow nd Fn Hed Dt 5 Focusing on norml operting conditions fns operting in the stble region, nd from 5 to % of the design flow rte leves totl of 47 dt in this study. Figure 6b compres the stble opertion zone mesured dt to the fn irflow sttion modeled dt. The fn irflow sttion vlues mtch the directly mesured vlues very closely. A liner regression of the dt with the regression forced through zero, gives n error of.7% for the cubic model, nd.5% for the qudrtic model. When the regression is not forced through zero, the y-intercept vlues re less thn 6 cfm (. m³/s), which is lower thn the mximum mesurement error. Thus the nd nd rd order models give nerly identicl ccurcy. The dt generted in the experiment pper to be highly repetble. There re no drstic outliers, or unexpected results within the norml fn opertion rnge. The very close greement between mesured nd predicted vlues shows tht Pressure cross fn (P) Fn Hed (P) 4

6 the irflow cn be ccurtely predicted using fn speed nd fn hed s inputs. The proposed method hs excellent potentil for ccurte determintion of irflow in ctul systems becuse the fn speed nd fn hed mesurement methods in ctul systems re very similr to those in the experiment. Theoreticl Flow (cfm) Theoreticl Flow (cfm) Mesured Flow (m³/s) Mesured Flow (cfm) () Mesured Flow (m³/s) y =.5x R = Mesured Flow (cfm) (b) Figure 6. Comprison of Mesured Airflow to Predicted Airflow Using the Second Order Model Theoreticl Flow (m³/s) Theoreticl Flow (m³/s) experiments could exmine the effects of pressure trnsducer loction, use of mnufcturer s fn curves rther thn in situ dt, nd the performnce of the fn irflow sttion model on belt driven fns. NOMENCLATURE H fn hed N fn speed Q irflow rte i polynomil regression coefficients ω fn speed rtio Subscripts d design, t % fn speed r return s supply xf exfiltrtion xh exhust APPENDIX: ERROR ANALYSIS The fn irflow sttion use the mesured fn speed nd fn hed to predict irflow, using the following eqution: H 4 ω ω ± Q = (A-) The mximum irflow prediction error depends on the ccurcy of both the fn hed nd the fn speed mesurement. Q Q Q = δω + δh ω H δ (A-) By introducing Eq. (A-) into Eq. (A-), the mximum irflow projection error becomes: CONCLUSIONS The fn irflow sttion model vlues mtch the direct mesured irflows well nd show excellent repetbility. This leds to the conclusion tht the fn irflow sttion model is vlid. A comprison of the results using the qudrtic nd cubic models shows no dvntge to the cubic model. The cubic model is difficult to implement due to its complicted solution. Since the fn curve cn be represented ccurtely in the norml operting rnge using the second order regression, this model should be used. Becuse fn irflow cn be mesured ccurtely using fn speed nd fn hed, the fn irflow sttion provides n ccurte nd cost effective mens of irflow mesurement in HVAC systems. A direct nd importnt ppliction of this model is building pressuriztion control. These experiments were conducted using direct drive plenum fn under controlled test conditions. Further (( ) ) 8 δq = Q + ω δh + ( δω (A-) Q + ω The mximum mesurement errors of fn speed nd fn hed re rpm nd.5 in w.g. (. P), respectively. The regression coefficients nd re -.9 nd.8, respectively. Figure A- shows the mximum predicted irflow error s function of irflow nd fn speed. The mximum error decreses from cfm (9.4 x -4 m³/s) to. cfm (4.7 x -5 m³/s) when the irflow increses from 5 cfm (.4 m³/s) to 5 cfm (.m³/s). The mximum reltive error is less thn.5%. 5

7 Mximum Error (cfm) RPM=5 RPM= RPM= RPM=5 Airflow Rte (m³/s) Airflow Rte (cfm) Figure A-. Mximum Projected Airflow Error versus the Airflow nd Fn Speed REFERENCES [] Brothers, P.W., Wrren, M.L., 986, Fn Energy Use in Vrible Air Volume Systems, ASHRAE Trnsctions, 9:B, 9-9. [] Lorenzetti, D.M. nd Norford, L.K., 99, Mesured Energy Consumption of Vrible Air Volume Fns under Inlet Vne nd Vrible Speed Drive Control, ASHRAE Trnsctions, 98:, [] Dmino, L. nd Dougn, D.,, Controlling Mold Growth in Exterior Wlls of Buildings by Continuously Mximum Error (m³/s) Mintining Positive Pressure Flow through Dynmic Ventiltion Strtegies, Proc. EPA/A&WMA Symposium. [4] Shkun, W., 99, The Cuses nd Control of Mold nd Mildew in Hot nd Humid Climtes, ASHRAE Trnsctions, 98:, 8-9. [5] Dmino, L., Dougn, D., nd Solberg, D.W., 99, Mesurement for the Control of Fresh Air Intke, ASHRAE Journl, :, [6] Elovitz, D.M., 995, Minimum Outside Air Control Methods for VAV Systems, ASHRAE Trnsctions, :, 6-8. [7] Hll, J.D., Mudrri, D., nd Werling, E, 996, Energy Cost nd IAQ Performnce of Ventiltion Systems nd Controls, Proc. Pths to Better Building Environments Bltimore, ASHRAE IAQ Conference, Teichmn, KY, 5-6. [8] Tylor, S.T.,, Compring Economizer Relief Systems, ASHRAE Journl, 4:9, -4. [9] Cohen, T., 994, Providing Constnt Ventiltion in Vrible Air Volume Systems, ASHRAE Journl, 6:5, 8-4. [] Kettler, J.P., 995, Minimum Ventiltion Control for VAV Sytems Fn Trcking vs. Workble Solutions, ASHRAE Trnsctions, :, 65-. [] NEBB, 986, Testing Adjusting Blncing Mnul for Technicins, Ntionl Environmentl Blncing Bureu. [] Liu, M, Vrible Speed Volumetric Trcking for Airflow Control in Vrible Air Volume Systems, Accepted by the Journl of Solr Energy Engineering. 6

TECHNICAL INSIGHT INTERNAL CLEARANCE - TYPES AND NORMS

TECHNICAL INSIGHT INTERNAL CLEARANCE - TYPES AND NORMS TECHNICAL INSIGHT PRODUCT AND APPLICATION ENGINEERING INFORMATION A PUBLICATION OF NSK AMERICAS INTERNAL CLEARANCE - TYPES AND NORMS Internl clernce is the distnce which the two rings of non-instlled bering