VOL. 8():99-9 99 Amerian Soietv of Agriultural Engineers 000-^^ / 0 / ^Rm-OO^o DIAGNOSTI HARDWARE/SOFTWARE SYSTEM FOR ENVIRONMENT ONTROLLERS K. hao, R. S. Gates, H.-. hi ABSTRAT. A system onsisting of a ommerial environment ontroller attahed to a personal omputer programmed to exeute a dynami simulation model for broiler house thermal response was designed. The building model updates interior environment from solution of first order heat and moisture balanes. Feedbak of predited interior temperature is provided to the ontroller; heating and ventilation equipment assigned in the model affet heat and moisture balanes and hene the system response. An objet oriented design approah was utilized and desribed. Governing equations for the building and ontroller temperature iruit dynami response were solved in real time with a series of simulations to assess the response of the interior environment to different ontroller settings. Assessments of ontroller stage differential, building minimum ventilation rate and system sampling time are provided for a simulated broiler house. The ombination of an environment ontroller and a building thermal model suh as that presented here an be used to evaluate environment ontrollers and building heating and ventilating design, as well as for training purposes. Keywords. omputer simulation. Temperature ontrol. Thermal environment. Poultry housing, Psyhrometris, Studies and researh appliations involving miroproessor-based environment ontrol have been performed by many researhers in a variety of areas inluding appliation of ontrol strategies to the plant or animal housing (ole, 980; Mithell, 98; Jones et al., 98), miroproessor-based optimal prodution systems (Timmons and Gates, 98; Gates and Timmons, 988), and integrated environmental ontrol systems (Gates and Overhults, 99). These efforts begin by larifying the onepts of miroproessor-based environment ontrol, refining environment ontrol strategies, and gradually leading to eonomi optimization, the ultimate objetive of environment ontrol. A primary requirement for operating ontrolled environment agriultural prodution failities is to manipulate the interation between the interior environment and the oupant produtivity (assuming adequate oupant health, geneti potential, and nutrition). To suessfully manipulate the interior environment, the operator (human or ontroller) must fully omprehend the interations among the heating system, temperature and humidity levels, and the operation of the ventilation system. The operator must be able to ontrol these interations in suh a way as to maintain onditions whih result in maximum net return (Timmons and Gates, 98). To understand suh a ompliated proess, a dynami Artile was submitted for publiation in May 99; reviewed and approved for publiation by the Information and Eletrial Tehnologies Div. of ASAE in January 99. Presented as ASAE Paper No. 9-0. This work is published with the approval of the Diretor of the Kentuky Agriultural Experiment Station and designated Paper No. RIS 9-. The mention of produt names is not an endorsement of the equipment but merely serves for doumentation purposes. The authors are Kuanglin hao, ASAE Student Member, Graduate Student, and Rihard S. Gates, ASAE Member Engineer, Assoiate Professor, Dept. of Agriultural Engineering, University of Kentuky, Lexington; and Hsien-hung hi, Software Engineer, ummins Engine Laboratories, olumbus, Ind. simulation model is helpful. While it is theoretially possible to ahieve very tight environment ontrol, this may require large energy inputs when desired setpoints annot be ahieved by simply ventilating. For example, while it may be straightforward to provide a uniform 0*^ environment during muh of the year in livestok buildings, to maintain tight ontrol around this temperature setpoint in very old onditions, or for outside onditions above 0, requires both proper apaity heating and ventilating equipment, and for the ase of hot weather a suitable form of supplemental ooling suh as heat pumps. onsequently, improvement of agriultural environment ontrol systems is generally onstrained to allow rather broad daily and seasonal flutuations in interior environment. Methods to further redue these flutuations by enhaned ontroller design must aount for the types and apaities of heating and ooling equipment being used, the building oupants (i.e., livestok, food or nonfood rops), and an aeptable level of energy use. One approah that has been used suessfully has been to simulate the entire environment ontrol system, inluding the building thermal environment, the ontrol system, and the building oupants. Zhang and Barber (99) have ompared various tehniques of thermostat and humidistat onfiguration strategies via simulation tehniques. They found that thermostat ontrol of ventilation and humidistat ontrol of heating during old weather resulted in improved ontrol, but the method whih used less heating energy resulted in lak of humidity ontrol- Berkmans et al. (99) utilized a dynami simulation to develop a simulated ontrol system and predited a 9% redution in annual energy onsumption ompared to a referene system. This full simulation approah to ontroller design annot suessfully mimi the behavior of many existing environment ontrollers, beause the atual dynami model for the ontroller is unknown. Miroproessor-based ontrollers utilize sophistiated digital filtering tehniques for measured data. Transations of the ASAE
and may atuate ontrol responses based on logi that inludes long-term history suh as degree of heat stress (Timmons et al., 99). A system that diretly inorporates an existing ontroller into a dynami simulation ould overome this limitation. The objetive of this researh was to develop a prototype system, onsisting of a miroproessor-based environment ontroller onneted to a personal omputer (P) running a dynami simulation model of a given broiler house. This system will provide a simulated design environment to investigate the dynami response of the interation between ontroller, sensor, building, and environment. For a speifi building heating and ventilation onfiguration the ontroller's logi and performane an be observed in real time. The system an be used as a training tool for users of a ontroller in various building onfigurations, as a omputer-aided-design tool for ontroller design engineers, or as a means of improving a given environment ontrol design. DYNAMI MODEL A omputer model based on lumped thermal energy balane equations was applied to simulate the dynami response of dry-bulb temperature and humidity ratio (kilogram water/kilogram dry air) inside a typial modem broiler house. The sensible and latent heat balane equations are: Equations and are linear first-order differential equations with variable oeffiients whih an be solved numerially using finite differene or a Runge-Kutta tehnique. It is instrutive to rewrite equations and as differene equations to examine effets of the oeffiients: where 'x«n _ nn- ^b+ts ^b+ts, ^bts (Qs+Qe+Qb+Qf ppv ^b+t, wr lb wr"'+ ^^ w ^b+t^ tb+ts TKT, b^s T,+ T,\pVhj (UA + FP)+X () () (),dt pv^-m(t -T.)+Q,+ Qb+Qe+Qf (D dt and? V (). dw pvhfg^=mhjw,-w.)+q, dt () All symbols are defined at the end of the artile. Assumptions used to derived these equations are:. Air inside the building is well mixed so that temperature is independent of spatial oordinates, T(x, y, z, t) «T(t) and an be measured by a single. sensor in one loation. The size and number of fans running at a speifi instane are assigned using ombined equipmentto-relay mappings, thus air flow rate an be determined by reading the ontroller's relays. Bird sensible and latent heat prodution are evaluated as funtions of bird age and mean room temperature (Gates et al, 99,99). hanges in heating and ventilation swithing are modeled as ramp disturbanes over a short time, rather than step disturbanes.. Building thermal storage other than that in the air is negleted. Broiler heat prodution equations (Gates et al., 99, 99) were used for Q^ and Qj. The strutural heat loss and heat exhange through the floor were alulated from: Q,«(:UA + FP)(T.-Tj () The building time onstant x^ for temperature response varies tremendously due to the ventilation term, x^, whih varies between about to 00 s. By ontrast the first term in the right hand side of equation is approximately. s for the simulated broiler house and its ontribution an be onsidered negligible. Thus, for a broiler house a single time onstant x^ is adequate to desribe transient response. However, other buildings with different ventilation apaities or signifiant thermal mass may not be realistially modeled by equations and. To obtain a numerial solution to equations and, it is important to selet a proper sampling time and integration time step, Tg. This sampling time is how often the ontroller is sampled by the simulation program to determine what equipment is ativated, and how often the ontroller temperature iruit is provided with an updated inside temperature. A useful way to haraterize the sampling time is to define a variable that has a reasonable physial interpretation. For first-order systems, the rise time is a natural normalization fator. The rise time is defined as the time required for the step response to rise from 0 to 90% of its final value. In order to reonstrut the ontinuous time signal fidelity, we estimated minimum system rise time as the building time onstant expressed in equation. Then the number of sampling periods per rise time, N = Tj / Tg an be defined. The minimum building time onstant is about s at full ventilation. From ontrol 90 TRANSATIONS OF THE ASAE
\//^T /\.Q'JO QAI engineering experiene, it is reasonable to hoose N between and 0, whih results in a time step of to s. MATERIAL AND METHODS The struture of a dynami simulation system for environment ontrol (fig. ) ontains an environment ontroller, an analog-to-digital onverter (AD), building dynami model, and a digital-to-analog onverter (DA) with zero order hold (ZOH) to model the temperature sensor. For a speifi building onfiguration, the building dynami model evaluates building ambient temperature and humidity ratio in real time. The relative humidity at the sampling instant is alulated by the omputerized psyhrometri routines (Zhang, 99). The building temperature is then onverted into an analog signal using the DA. This signal is maintained onstant between the sampling instants by a ZOH, and sent to the ontroller temperature iruit. The measured error between the setpoint temperature and the output temperature of the building dynami system is used by the ontroller to make hanges in ontrol relay settings as depited by the staging diagram (fig. ). The relay states are read by the program and used as inputs (V and Qf) to the building dynami model (i.e., eqs. and ) and the solution is used to update building temperature and humidity ratio. The primary funtions for this system are disussed below. ONTROLLER A miroontroller-based environment ontroller (SE, model Aerostager, Aeroteh In., Mason, Mih.) was used in this researh, although any environment ontroller ould be used. The SE ontrols temperature in a two-step proess (Gates and Overhults, 99):. The feedbak error between measured inside temperature and the setpoint temperature is used as input to a ventilation program to determine what stage of ventilation is ative.. The urrent stage is used as input to a stage program to determine whih relays are ative. Preprogrammed ventilation and staging programs were seleted as shown in tables and. A sequentially staged relay atuation was used as depited in figure. In addition, the SE provides a 0-min interval timer ontrol yle for minimum ventilation stage (fig. ). If the urrent building temperature is less than or equal to the setpoint temperature, then the yle on-time is as seleted by the user. If, however, the building temperature exeeds the setpoint temperature, then the yle on-time inreases proportionately with feedbak error over the stage interval. Minimum Ventiiatioa Heat dt t dt Setpoint dt dt dt dt dt dt Figure - diagram for SE. Note: Uniform staging diffierential (dt) between stages with a dt heat offset SIMULATION OF DYNAMI SYSTEM The systemati representation of the dynami proess of a building environmental ontrol system an be expressed in three basi ategories, i.e., input, system dynami model, and output. Figure is a blok diagram illustrating this proess. Input to the dynami simulation system inludes ontrol variables, building and ventilation parameters, and disturbanes. ontrol variables are ventilation and supplemental heat rates whih are atuated by the ontroller. Parameters are stati values (e.g., heating and ventilation apaity, volume of building, et.) and/or slow time-variant variables (e.g., overall building sensible heat transfer oeffiient, age of bird, and numbers of birds, et.) whih are not expeted to hange rapidly one a design is formulated. Disturbanes, on the other hand, are dynami values whih are strongly time-dependent; for example, the ativation of a fan or heater. Both ontrol variables and disturbanes originate from the ontroller and are affeted by temperature feedbak. Thus, as simulated temperature hanges the SE will ativate a different stage and this hange will at as a disturbane. The "equipment" onneted to the ontroller was speified as part of the building model. When a hange was sensed in relay states a ramp was initiated to apportion the hange in heating or ventilation energy over this period. A ramp interval of s was hosen. DATA AQUISITION PROESSES The system dynami model onsists of a systemati appliation of basi physial law, i.e., sensible and latent heat balane equations. Output of the simulated proess is inside building temperature and relative humidity. The interfaes between system's input and output are Table. Example staging program for the ontroller Setpoint ontruiiei A AD ontrol Variable Temperature Sensor Model DA+ZOH Building Dynami Model Tpredi Figure -Struture of dynami simulation system for environment ontrol. Level Heat 8 Differential TF) (offset) Lower Hysteresis ( F) N/A Upper Hysteresis ( F)
Table. Example ventflation program for the ontroller Relays ( - losed relay, MT - Modulation relay, time yle) Rl R R R R R R R8 R9 Heat MT MT represented by data aquisition proesses. The funtions of these proesses are disussed below. AD. A data aquisition board [DASON- (Keithley Metrabyte, 0 Myles Standish Blvd., Taunton, MA 080)] was used to interfae the ontroller to the P. The ontroller omputes the error signal (differene between the setpoint and output of the dynami proess) and uses the staging ontrol algorithms to generate a new relay atuation. An on board general purpose programmable I/O devie, Intel 8A, was used to determine equipment status. The simulation proess read the status of port B (bit 0 to bit ) and port (bit 0 to bit ) of an Intel 8A, whih was onneted to tfie ontroller's ontrol relays to represent the equipment (fans and heaters). The assigned values of ventilation rate and supplementary heat in the simulation program were used for the next integration time step. DA + ZOH. The temperature sensor [LM (National Semiondutor, 900 Semiondutor Drive, Santa lara, A 90)] for the SE produes an analog signal linear to the temperature. The SE samples the sensor several times per seond, aumulating these disrete samples. The temperature displayed on the SE is used for ontrol purposes and is omputed from its aumulated disrete samples approximately every 8 s. Additional ontrol logi inluding user-speified hysteresis between stages determines whih relays are ativated. To provide temperature sensor inputs to mimi those of the SE, a model for sensor response was onstruted and ombined with a hold element to reonstrut the analog signal. First the temperature sensor time onstant (Xg) was determined experimentally. Then a first-order with zeroorder hold model was onstruted to simulate the response of SE's temperature sensor. The differential equation desribing a first-order devie is: ontrol Variables Ventilation Rate Supplementary Heat ^ Building Data Init Bldg Air Temp and Hum Init Outside Temp and Hum Heat and Ventilation apaity Volume of Building UA+HF of Building Numbers of Birds Age of Bird Dynami Model Sensible and Latent Heat Eqs Sensor Simulation Building Air Temp and Humidity Figure -Proess for dynami simulation of environment ontrol system. ln(l-y)«ln l Id_JIaL = L (0) T.-T; \ The overall time onstant of the sensor and ontroller software an be determined by a regression method to equation 0; a value for x^ of approximately s was obtained for a step hange from 0 to. To model the temperature sensor response, taking the Laplae-transform on both sides of equation 8, and dividing the output (Y) by the input (X), the impulse transfer funtion yields: G.(s)-X(!).^^ X(s) v+ () Equation an be simplified using a = l/tg and it leads to: G,(s) = K () s+ a A hold element H(s) is added to the impulse transfer funtion Gi(s) to develop the sensor impulse transfer funtion: G(s ) = H(s)G.(s)=^-"^P^-^'^^) Ka^ () s+ a (T,D + l)y(t) = Kx(t) (8) Solving equation 8, when subjeted to a step hange with zero initial onditions, and dividing by the magnitude of the step hange Kx leads to: J^=l- Kx exp fe)' T. - T.; where the right-hand side of equation 9 is represents the relative hange in displayed temperature sine the step hange was invoked. Letting Y «- exp(-t/ts), substituting into equation 9, and taking logarithms yield: (9) To disretize the sensor's transfer funtion G(s), a Z-transform of both sides of equation and leads to: G(z). [l-exp(-atj]k_ z-exp (-TJ bz'' + az"^ () The system transfer funtion G(z) is then reonstruted to a first-order differene equation by taking inverse Z-transform of equation, and it follows that: yn = -ayn-l+bxn_i () 9 TRANSATIONS OF THE ASAE
VOL. 8():99-9 9 Substituting K =, x^ = s, and T = s yields a = -0.9 and b = 0.0. Thus the sensor output is a weighted sum of the sensor's previous output and the previous temperature from the building thermal model. DESIGN OF SIMULATIONS Software development for this diagnosti system was based on objet oriented design (OOD). The designed OOD model onsists of four layers: lass-and-objet, Struture, Attribute, and Servie. This four-layer struture sheme is illustrated in figure. We defined Building as a lass with instanes of Building Data and Building Model. The definition of this OOD model reflets the designed system with apabilities of enapsulation of attribute values and their exlusive servies (oad and Yourdon, 99). A Whole-Part struture suh as illustrated in figure is used to express the problem domain omplexity, relevant to the system's responsibility (i.e., reation to interations between building, environment, and ontroller). It is shown with a whole objet (Building) at the top, and then individual part objets (ontroller, Sensor) below, with a line onneted between them. The Attribute setion of an OOD model is state information for whih eah objet in a lass has its own value (e.g., attributes of PsyBase suh as relative humidity, dry-bulb temperature, humidity ratio, et.). In order to fulfill its (objet) responsibilities, an instane onnetion is needed to map one objet to other objets. For example, instane onnetions of objet Building to objets PsyBase and BirdHeatProdution were made to evaluate building air state properties and bird sensible and latent heat produtions within building. Message onnetions are servies between objets and also indiate a need for servies for a Whole-Part struture to perform its responsibilities. For example, the Sensor objet sends a message to the ontroller objet to omplete the feedbak loop. The notation for a message onnet is an arrow pointed from sender to reeiver (the psyhrometri routines, in ++, are available from the authors upon request). ontroller Air Infiltration Ventilation Heat Initialize Sample AD DA ' Building Building Data Building Model '^. ^" ~" P \ L Sensor Time onstant Initialize Sample ^ J L Z, PsyBase Dry-Bulb Temp Relative Hum Water Yap Pres Part Pressure Humidity Ratio Dew Point Temp Enthalpy Wet-Bulb Temp BirdHeatProd SensibleHeatProd LatentHeatProd Figure -Objet oriented design representation for building environment ontrol system. PROTOTYPE EVALUATION A ommerial broiler faility with parameters speified in table was used. An infiltration rate of one air hange per hour was assumed. Heaters were speified to be 0 kw LPG, and timer fans were speified to be. m^/s (i.e., nominal -in. fans). Higher stage fans were larger (i.e., 8 in.), with two fans per stage, 8.9 m^/s. Other building details are similar to those given in Gates and Overhults, 99. Two sets of example onditions were investigated to assess the prototype's behavior (table ). The first set onsisted of simulating young birds (-0 days) with old outside temperatures, suh as expeted during winter brooding. Two different stage differentials ( or F) and various ombinations of heaters and timer fans were simulated in five separate runs. The seond set onsisted of simulating 0-day old birds during mild onditions ( ) suh as experiened during bird growout. The effet of sampling time was determined, and the stage differential was hanged, for a total of three separate runs. RESULTS AND DISUSSION To determine what equipment is ativated from the program output, a Relay State variable was defined and its values are given in table. States 0 and reflet the two possible ases when heat is used either the timer fan is off (state 0) or it is on (state ). States and orrespond to stage in figure ; depending on the timer yle the minimum ventilation fans are either off or on. Relay Inside RH 0% Inside temp Outside RH 0% Outside temp ontrol setpoint 0,000 birds 0 days of age / house brooding Heat loss: 0 W/K Table. Simulation performed for study Initial onditions Bird onditions Test Runs Inside RH % Inside temp Outside RH 0% Outside tempi ontrol setpoint 0,000 birds 0 days of age Full house Heat loss: 88W/K asel ase : Ts - s, stage differential F min on-time for timer timer fan, heaters : Ts - s, stage differential F min on-time for timer timer fan, heaters : Ts - s, stage differential F min on-time for timer : Ts - s, stage differential F min on-time for timer : Ts - s, stage differential F min on-time timer : Ts - s, stage differential F min on-time for timer : %"! s, stage differential F min on-time for timer : Ts - s, stage differential F min on-time for timer
Table. Relay state definitions Relay State Heat Ventilation 0 8 9 On On H H Infiltration only Minimum ventilation fans On Intiltration only Minimum ventilation fans On ventilation ventilation ventilation ventilation ventilation ventilation States and higher reflet straightforward appliation of additional fans. SIMULATION ASE Figure is an example of the type of plots onstruted from the simulations. In figure, the upper plot is the Relay State time history, the middle plot shows predited and sensor temperature traes, and the lower plot shows predited temperature and relative humidity. Initial onditions for all ase runs were lose to the ontroller setpoint so that no signifiant initial disturbane was reated. The plot of temperatures in figure appear to be in steady periodi equilibrium, varying from about. to, with a period of approximately 00 s. The sharp nature of the building temperature is due to the lak of any heat storage terms in the first order equations that were used to model the sensible and latent heat balanes. The smoother trae of the simulated sensor output does not reah the magnitude in peaks that the building temperature ahieves, due to its dynami response harateristis. Inspetion of the output traes shows that peak building temperatures vary somewhat due to ativation of the 0-min timer fan for min out of eah 0-min yle (orresponding to relay state value of or ). The ontroller begins in relay state (heat off, fans on), but quikly hanges to state when the timer fan shuts off. The relay state then shifts to state 0 beause the temperature has dropped below the heater setpoint. This auses the building temperature to raise and the relay state returns to state (heater off, infiltration only). The timer ativates at about S.Vlrirti ' n ' n /' \ /I \ / ' -^ ^ r- J\ / \ / V \ RHprBdit 90 ^ ^' 80 = E 0 0 > 0 000 000 000 000 000 000 Time (s) Figure -Simulatioii results for ase, run : one timer fan, -min on-time, ** F stage differential. 00 s into the simulation and runs for min, ausing the temperature to drop and the heater to ativate again. One the timer shuts off, the heater remains on until the building warms up. The building then repeats the yli temperature variation, ooling off slowly at first due to infiltration and then more rapidly when the timer fan turns on. Relative humidity requires muh more time to equilibrate to steady periodi behavior. It yles between to 80% after about 00 s. Run (fig. ) was similar to run exept that the timer was speified at min on-time rather than min. The effet on the system was that building minimum temperature inreased slightly and relative humidity reahed saturation after about 000 s, indiating that the minimum ventilation was insuffiient for this ase. Run (fig. ) was idential to run, exept that two timer fans were used to alleviate saturation onditions. Building minimum temperature was redued to about. and relative humidity osillates between about 8 to 98%. ompared with run, there is little differene in B is VW^^^VVV\r JS ^ 0 ^j^v^^i^^mrir I " S. E «O).E m 0 h A A A A /irhpoidit - 000 000 Time (s) ^ 0? HBO a, 0 ^ a> s S. E a>.e m 0 A A l\a/\/vvv\ 80 000 000 000 000 000 000 Time (s) ^ E 0 0 I 0 Figure S-Simulation results for ase, run : one timer fan, -min on-time, F stage differential. Figure -Simulation results for ase, run : two timer fans, -min on-time, ^ F stage differential. 9 TRANSATIONS OF THE ASAE
: ^ VOL. 8():99-9 9 O 0 "^predit ~ Tpy^it *^ \ / ^ A/ V/ 0 00 000 00 000 00 000 Time (s) 0 = 0 I Figure S-Simulation results for ase, run : two timer fans, -min on-time, F stage differential. temperature and relative humidity between one fan for min versus two fans for min. For run (fig. 8) the ontroller ooling stage differential was redued from to ^ F and two timer fans were run for min out of 0. The lower temperature during eah osillation was about. older than in run, owing to the doubling of fan on-time. Maximum temperature was similar to run. The sequening of equipment was also different beause the timer was off during the first 00 s of the run. Relative humidity was redued to to % due to the additional minimum ventilation provided. Run was idential to run exept that a F stage differential was used. The temperature and relative humidity plots (not shown) were very similar to those from run, indiating minimal influene of stage differential on the frequeny and magnitude of osillations. SIMULATION ASE ase tests were seleted to investigate system response in the ventilation ooling mode. The birds were older (0 days) and mild outside onditions were used. Initial inside temperature was set to and the ontroller setpoint was set to 0, ausing a large initial step disturbane to the system. Only the first 00 s were reorded. Runs and utilized min on-time, two timer fans and a stage differential of and F, respetively. Relay states, predited building temperature, and sensor temperature are plotted in figures 9 and 0. After ativation of high ventilation to redue the initial high temperature, both tests spent most time in relay states and (ooling stages and ). Run took about 0 s to eliminate the initial disturbane and run required about 00 s. The temperature during run was about ooler than run and on two instanes the relay state dropped to stage (infiltration only). Relative humidity in run was to % greater than in run due to the slightly greater ventilation; both runs ahieved relative humidities of 0 to %. Run was idential to run exept that a -s time step was used. Dynami response to the initial disturbane was slightly damped, indiating some adverse numerial effets Figure 9-Simulation results for ase, run : two timer fans, -min on-time, ** F stage differential. from the larger time step. Longer term equilibrium temperatures were very similar and entered between about and for both tests. Relative humidity, however, was signifiantly different between the two runs. For run it was about to % and for run it was about 9 to %. These differenes an be attributed in part to the fat that the ontroller updates its ontrol relays approximately every 8 s, so that when sampling at s some ontrol intervals are missed. For example, in 00 s there will be 80 sampling points and 0 ontrol periods; when sampling at s there are 00 sampling points. The larger time step auses the simulated building system to miss some energy inputs and results in an artifiially dampened response. The system appears to operate reasonably for the onditions explored. This approah offers several advantages over other straight-forward simulations in the literature whih inorporate both the building and the ontroller into the simulation. A primary benefit of this approah is that a speifi ontroller, whether a ommerially available unit or a researh prototype, an be evaluated. This is espeially important for evaluating system response of digital ontrollers, beause the system response will depend on the digital filtering tehniques employed in the ontroller software. It is ertainly possible to simulate a ontroller's behavior, inluding suh features as sampling frequeny, sample averaging, ontrol interval and subinterval lengths, et., and inorporate this model Figure 0-Simulation results for ase, run : two timer fans, -min on-time, F stage differential.
into a building system simulation (assuming ooperation on the part of the manufaturer to divulge their algorithm). However, the method utilized in this researh allows the ontroller to be treated initially as a "blak box" of unknown response. This may be helpful for evaluating a partiular ontroller, or an environment ontrol system design, to estimate the effet of hanging various ontroller options suh as stage differential or hysteresis, or to estimate the effet of plaing environment ontrol equipment on different stages of a ontroller. We have used the system in graduate teahing laboratories to demonstrate the use of a P-based simulator as part of a ontrol system evaluator; and have used it to evaluate alternative ontroller strategies suh as the use of running average temperatures and relative humidities as setpoints (Timmons etal., 99). One onstraint to this approah is that the simulations must be arried out in real time. While the building thermal model is not onstrained to real time operation, the environment ontrollers are designed to operate in real time and hene an evaluation of the ontroller's effet on a system must also be performed in real time. Also, the ontroller's temperature iruit must be haraterized and the proper signal sent from the P to the ontroller via DA. Additional researh on this system to further determine the building thermal model by omparison with field data would be useful. To extend tfie approah, it may also be neessary to improve the diagnosti system's adequay by inluding heat and moisture storage terms, as the thermal apaitane of a struture and the ability of floor litter to absorb some moisture on a short-term basis an signifiantly affet the dynami balanes. A similar analysis, applied to greenhouse and swine failities that exhibit more thermal mass and solar radiation effets would be of interest. ONLUSIONS Based on the work reported in this artile the following onlusions an be made: The diagnosti system developed in this work appears to reasonably mimi expeted building response. Simulated interior temperature and relative humidity are used as feedbak to a realtime ontroller whose ontrol relays are monitored for heating and ventilation rates. Seletion of sampling and integration time step of s slightly dampened temperature response and signifiantly affeted relative humidity response ompared with -s time step, for the ontroller used in this study. The use of an objet oriented design allows onsiderable flexibility for future appliations to other livestok failities and environment ontrollers, through the use of lass-and-objet onepts applied to speifi implementations. The developed system has proven useful for both teahing and for investigating alternate ontrol strategies. AKNOWLEDGMENT. This work was part of USDA-SRS Regional Projet S-, "Systems for Providing and ontrolling Interior Environments for Poultry and Livestok". REFERENES Berkmans, D., E. Vranken and M. Van Pee. 99. Analysis of the ontrol of livestok environment by simulation tehnique and field data. ASAE Paper No. 9-. St. Joseph, Mih.: ASAE. Goad, P. and E. Yourdon. 99. Objet-Oriented Analysis, nd Ed. Englewood liffs, N.J.: Prentie-Hall. ole, G. W. 980. The appliation of ontrol system theory to the analysis of ventilated animal housing environments. Transations of the ASAE (): -. Jones, R, J. W. Jones, L. H. Allen Jr. and J. W. Mishoe. 98. Dynami omputer ontrol of losed environmental plant hambers: Design and verifiation. Transations of the ASAE ():89-888. Gates, R. S. and M. B. Timmons. 988. Miroproessor ontrolled broiler environment for optimal prodution. In Pro. Third Int. Livestok Environment Symp., -. Toronto, - April. St. Joseph, Mih.: ASAE. Gates, R. S. and D. G. Overhults. 99. Field evaluation of integrated environmental ontrollers. ASAE Paper No. 9-0. St. Joseph, Mih.: ASAE. Gates, R. S., S. H. Zhang and D. G. Overhults. 99. Minimum ventilation for broiler housing. ASAE Paper No. 9-. St. Joseph, Mih.: ASAE. Gates, R. S., D. G. Overhults and S. H. Zhang. 99. Heat and moisture prodution for modem broilers. In Pro. Fourth Int. Livestok Environment Symp., -8. England, -9 July. St. Joseph, Mih.: ASAE. Timmons, M. B. and R. S. Gates. 98. Eonomi optimization of broiler prodution. Transations of the ASAE 9(): -8, 8. Timmons, M. B., R. S. Gates, R. W. Botther, T. A. arter, J. T. Brake and M. J. Wmeland. 99. TIV algorithms for poultry environmental ontrol. ASAE Paper No. 9-08. St. Joseph, Mih.: ASAE. Mithell, B. W. 98. Miroomputer-based environmental ontrol system for a disease-free poultry house. Transations of the AA :9():-0. Zhang, S. H. 99. Minimum ventilation for broiler housing. Unpub. M.S. thesis, Dept. of Agriultural Engineering, Univ. of Kentuky, Lexington. Zhang, Y. and E. M. Barber. 99. Variable ventilation rate ontrol below the heat-defiit temperature in old-limate livestok buildings. Transations of the ASAE (): -8. LIST OF SYMBOLS a -exp(-ats) a l/xg b K[l-exp(-aTs)] p speifi heat of air (00 J/kg- ) D differential operator FP heat loss through perimeter (W/ ) hfg latent heat of vaporization of water (.e J/kg) K stati sensitivity of sensor rii mass flow rate of air through the building (kg/s) N number of sampling points per rise time ^bird number of birds in building Ql heat loss from building due to ondution (W) Qe eletrial energy use (W) Qf supplemental fuel (W) Qs sensible heat generated by housed animals (W) Qi latent heat generated by housed animals (W) p air density (kg/m^) 9 TRANSATIONS OF THE ASAE
SUA overall building sensible heat transfer oeffiient (W/ ) T^i initial state atual temperature to sensor ( ) Tjj reading of measured temperature from ontroller Tj inside air dry-bulb temperature ( ) TQ outside air dry-bulb temperature ( ) Tj rise time [ s (minimum)] Tg sampling time (s) Tgs steady state temperature measured by sensor ( ) Tb time onstant of building sensible heat balane (s) T\y time onstant of building sensible heat balane due to the ventilation term (s) Tg time onstant of temperature sensor (s) Y building volume (m^) V volumetri flow rate (m^/s) Wbird bird weight (kg) Wj inside air humidity ratio (kg/kg) WQ outside air humidity ratio (kg/kg) x(t) atual temperature of sensor ( ) y(t) temperature measured by sensor ( ) VOL. 8():99-9 9