August 2010

American Society for Horticultural Science Volume 20, Number 4

August 2010

American Society for Horticultural Science Volume 20, Number 4

Virtual Grower —amanagementtoolforgreenhousegrowers

Virtual Grower —amanagementtoolforgreenhousegrowers

Virtual Grower: Software to Calculate HeatingCosts of Greenhouse Production in theUnited States

Jonathan M. Frantz1,4, Bryon Hand1, Lee Buckingham2, and

Somik Ghose3

ADDITIONAL INDEX WORDS. decision support, energy efficiency, floriculture, glazing,covering

SUMMARY. Greenhouses are used in many climates for season extension or year-round production and can be expensive to heat. Greenhouse users and growers areoften faced with management decisions that rely on an understanding of howtemperature settings, heating systems, fuel types, and construction decisionsinfluence overall heating costs. There are no easy-to-use programs to calculateheating costs associated with these factors over full cropping seasons. A computerprogram called Virtual Grower was created that helps calculate heating costs atmany U.S. sites. The program uses a weather database of typical hourly temperature,light, and wind information of 230 sites from the National Renewable EnergyLaboratory in the calculations. A user can define unique design characteristics suchas building material and construction style. The user also defines the type of heatingsystem and heating schedule, and then the program will predict heating costs basedon typical weather at the selected location. Shorter-term predictions with weatherforecasts of 2 days or less can be made with the software if there is an internetconnection through integration with local weather forecasts. Virtual Grower canserve as a platform from which many other features can be added, such as plantgrowth and scheduling. Continued development will improve the software andallow users to perform baseline analysis of their heating costs, identify areas in theirproduction to improve efficiency, and take some of the guesswork out of energyanalysis in unique greenhouses.

In many parts of the United States,plants are produced in green-houses in colder times of the year,

for early marketing of ornamentalcrops (Moccaldi and Runkle, 2007)or early/year-round vegetable pro-duction (Hochmuth and Hochmuth,1993). For this reason, energy costsare second only to labor costs as themost expensive factor in indirect costsof greenhouse production. While oiland natural gas prices fluctuated by100% or more in the last 3 years,generally, fuel prices have risen by

50% over the last 10 years (U.S. De-partment of Energy, 2009), and rep-resents a large cost that is oftendifficult to predict. Greenhouse ac-counting software has pooled thisnumber in variable cost estimates,typically on an area per time basis(Brumfield, 1992), or requires theuser to input this cost (Fisher andDonnelly, 2002). Thorough analysesor even a basic understanding offactors that contribute to this costare complex and can be difficult towork through on long-hand worksheet

calculations (Aldrich and Bartok, 1994;Brumfield, 1992). These worksheetsare excellent references in and of them-selves, but getting a grower or special-ized consultant to perform them can betime consuming, difficult, expensive, orall of these.

The National Renewable EnergyLaboratory (NREL) has a databaseof typical weather for 230 U.S. loca-tions, including hourly temperature,wind speed, and solar radiation (Marionand Urban, 1995). Combining thisresource with standard methods of cal-culating energy balance in a greenhousecould yield useful information aboutenergy requirements of different struc-tures and allow for simulations in dif-ferent locations and cropping periods.

A computer program was de-veloped called Virtual Grower thatcombines the NREL database withstandard energy calculations for green-house heating needs. The programalso provides methods of estimatingtypical commercial-scale heating sys-tem efficiencies and air infiltrationvalues. Flexibility in how a greenhouseis heated allows a user to estimatetypical heating costs associated witha variety of realistic heating schedulesor evaluate potential savings throughchanging methods, upgrading sys-tems, or switching to an alternativefuel source. Even though typicalhourly data are used in this program,accurate predictions of heating of agreenhouse have been achieved whenlonger-term (several month seasons,for example) time frames are simulated.

Program descriptionWhen the software is opened, the

main design page is viewed (Fig. 1).At the top of the page, there areclickable tabs labeled ‘‘Design,’’ ‘‘Heat-ing Schedule,’’ ‘‘Lighting,’’ ‘‘Costs,’’‘‘Plant Growth,’’ and ‘‘Real TimeData.’’ By clicking on the tabs, thosesections will appear. Note that thisarticle describes all of these tabs except‘‘Lighting’’ and ‘‘Plant Growth.’’

In the main design page, fourboxes are outlined: ‘‘Current Green-houses,’’ ‘‘Edit Greenhouse,’’ ‘‘FindSite,’’ and ‘‘Edit Site.’’ The ‘‘CurrentGreenhouses’’ box displays a list ofgreenhouses that have been createdwithin Virtual Grower. Upon open-ing the program, there are no green-houses in the list. When the button‘‘Add New Greenhouse’’ is clickedonce, a greenhouse name appears in

Technologyand Product

Reports

778 • August 2010 20(4)

this list next to a box that can bechecked or unchecked by clicking onthe greenhouse name in the ‘‘CurrentGreenhouses’’ list. This activates ordeactivates the greenhouse for a sim-ulation. A user can also load previouslydeveloped greenhouses from otherVirtual Grower sessions by clicking‘‘File-Load Settings,’’ and selectingfrom a list of saved greenhouses. Whena greenhouse is added to this list, the‘‘Total Square Feet’’ of all greenhouses(simulated and not simulated) isshown at the bottom left corner.

The user chooses a location inthe lower left side of the main page bychoosing the state and closest city orby searching for the nearest site bytyping in the zip code. This is an im-portant step that provides the weatherinformation for the core of the calcu-lations. Making a selection loads thehistorical weather information, includ-ing temperature, humidity, wind, andsolar radiation for that location. It isimportant to note that the weatherinformation is not average values fora period of time, but typical or repre-sentative values for that date and time.For example, a location might use datafrom Feb. 1987 followed by Mar. 1982because these months were deemedmost representative of the weather atthat location (Marion and Urban,1995). The value of this, as opposedto using averages, is that typical dailyfluctuations are simulated; if averageswere used instead, it would always be ablend of sunny, overcast, rainy, windy,calm, etc. A single day’s simulation,therefore, may not accurately predictthat day every year, but simulationsover longer periods of time would likelypredict that site with reasonable pre-cision yet show typical variability from

month to month or week to week thatcan be expected at that location.

The ‘‘Edit Greenhouse’’ box al-lows a user to name and define thesize and shape of that greenhouse.By clicking in each section (‘‘Name,’’‘‘Length,’’ ‘‘Width,’’ ‘‘KneewallHeight,’’ ‘‘Material,’’ ‘‘Fuel Type,’’and ‘‘Fuel Price’’), you can type inthe information specific for that green-house. Virtual Grower comes with amaterial database containing 15 com-monly encountered greenhouse mate-rials (Table 1) selectable through adrop-down menu with the materialnames for selection. If selected here,the greenhouse is constructed assum-ing that the entire greenhouse, otherthan the kneewall, is made of thatmaterial. Rarely are greenhouses builtwith such uniformity, so to refine thegreenhouse materials and other fea-tures, a user would click on the‘‘Advanced Design Options’’ buttondisplaying a different pop-up window(Fig. 2). Virtual Grower has a featurethat allows a user to enter uniquematerials with associated insulationand light-transmittance values allow-ing for testing of new, unique, exper-imental, or theoretical materials easily.

In addition to allowing for uniquecombinations of materials for differentwalls, roof types, and kneewalls, a usercan change the sidewall and centerheights, greenhouse style, and if agutter-connected greenhouse is beingconstructed, choose the number andwidth of spans within the structure.

At the bottom of this screen isa picture of the greenhouse style (rooftype) that is being constructed anda color-coded diagram that corre-lates with the different terms. A two-bay gutter-connected house is alwaysshown, but it does not mean all green-houses are gutter connected. There areseven roof styles that can be simulated,including Quonset-style greenhouses,

which can be built by setting the heightat the edge to zero and selecting theroof shape as ‘‘Arched.’’

Currently, each greenhouse istreated as a stand-alone structure withno shared walls with other houses orstructures.Thiswould influenceheat lossor gain and potentially have a significantimpact on the overall costs of heatingyour greenhouses (Giacomelli andRoberts, 1993; Simpkins et al., 1984).

Another option within the ‘‘EditGreenhouse’’ box is to define the airleakage or air infiltration of the green-house. In a careful energy audit,known amounts of trace gases suchas carbon dioxide (CO2) are addedto a greenhouse and the ‘‘decay’’ rateor disappearance of these gases ismeasured to compute leakage ina similar manner as a closed CO2 gasexchange system (Wheeler, 1992).Some common or typical values arereported in Aldrich and Bartok(1994) for new and old constructionmade of glass, fiberglass, or polyeth-ylene film. Using these values asa starting place, estimates were madefor commonly observed greenhouseconditions such as malfunctioning ormissing vent covers, gaps, or tears inthe glazing, and uninsulated parti-tions between greenhouse sections.Baseline conditions were simulatedfor greenhouses containing these con-ditions, and air infiltration rates werefurther modified so that the modeloutput matched with grower heatingbills. This empirical method of match-ing rough descriptors lacks the rigor ofa trace gas leak test, but allows a user toquickly describe their facility and ap-proximate their conditions.

In Virtual Grower, a user is al-lowed to select characteristics suchas approximate number of large andsmall gaps (with photographic ex-amples given), broken vent covers,and descriptions of the partitioning

UnitsTo convert U.S. to SI,multiply by U.S. unit SI unit

To convert SI to U.S.,multiply by

0.4047 acre(s) ha 2.47111.0551 Btu kJ 0.94783.6246 cord m3 0.27590.3048 ft m 3.28080.0929 ft2 m2 10.76390.0283 ft3 m3 35.31473.7854 gal L 0.26420.4536 lb kg 2.2046

105.4804 therm MJ 0.00950.9072 ton(s) Mg 1.1023

(�F – 32) O 1.8 �F �C (1.8 · �C) + 32

We thank the Toledo Area Flower and VegetableGrowers’ Association and Maumee Valley Growersfor helpful discussions in the development of thesoftware, and James Altland, James Locke, DouglasSturtz, and anonymous reviewers for helpful reviewcomments.

Mention of a trademark, proprietary product, orvendor does not constitute a guarantee or warrantyof the product by the U.S. Department of Agricultureand does not imply its approval to the exclusion ofother products or vendors that also may be suitable.

1United States Department of Agriculture, Agricul-tural Research Service, 2801 West Bancroft, Mail Stop604, Toledo, OH 43606

2University of California, Riverside, Department ofBotany and Plant Sciences, Riverside, CA 92521

3Center for Innovative Food Technology, 5555 Air-port Highway, Suite 100, Toledo, OH 43615

4Corresponding author. E-mail: jonathan.frantz@ars.usda.gov.

• August 2010 20(4) 779

between greenhouse areas. Each se-lection adds a known amount of airexchanges as an approximation of sys-tem leakage (Table 2). A tight, well-sealed greenhouse is considered tohave an air exchange rate of 0.5 airexchange per hour, while an extremelyleaky greenhouse has a rate of 5.1 airexchanges per hour. While this may

provide a user a better description oftheir system, a careful energy audit mayreveal a different estimate, which canbe entered directly by overriding thematrix-generated value in the lowercorner of this window. If nothing isselected in this section, the programhas default values of 1.3 air exchangesper hour, which is representative of a

medium-aged greenhouse with a fewholes throughout.

A description of the overall heat-ing system’s efficiency is also possible,and is set up in a similar manner as theair infiltration section. Again, a care-ful energy audit by qualified person-nel would result in a more accuratedescription of the heating systemefficiency. The method of assigningheating efficiency values was per-formed by measuring a number ofheating systems’ combustion and out-put efficiencies in situ. This informalsurvey comprised of several high-efficiency boilers, power-vented andgravity vented boilers, and unit heat-ers, all comprised of several ages andmaintenance schedules. The most un-certain aspect in the development ofthe decision matrix was delivery of theheat to the plants. Maintenance fre-quency was not weighted heavily be-cause there is considerable overlap inthe effect of maintaining a heater andour measured efficiency values in thesurvey. Finally, three to five growerheating bills were compared with base-line simulations to ensure that largererrors were accounted for in the matrix.

Working through the heatingefficiency matrix, a user can choosesimple descriptions of their system inVirtual Grower and help ‘‘dial in’’ anestimate for this important term. Bychoosing a combination of heater typesand ventilation, heat delivery method,and some description of age and main-tenance performed on the system (withphotographs to help a user makeappropriate selections), a single effi-ciency (from 0% to 100%) is calcu-lated. These three sections each havean efficiency value (Table 3) and aremultiplied together to estimate theoverall system efficiency. Similar tothe air infiltration section, a user canoverride the matrix-generated valueand enter a value from 0% to 100%.Some matrix combinations are notpossible. For example, if a user hasa standard unit heater, it is not possi-ble to select heated floors as a deliverymethod. If the user does not selectanything in this section, the programuses default values of 45%, which istypical of a forced-air system with noassistance to distribute that heat.

The user has the option of select-ing the fuel source for their heatingsystem from a drop-down list of 31fuel types (Table 4). These values areexpressed in terms of British thermal

Table 1. List of greenhouse materials in the database used by Virtual Grower,their unit area thermal conductance value (U-value), and light transmittanceproperties. Users can select from the different materials to construct a greenhouseand use different materials for different greenhouse walls. Values are based onthose from Aldrich and Bartok (1994). Note that U-value is the inverse of thethermal resistance value (R-value) commonly used in the building industry.

MaterialU-value

(Btu/h/ft2 per �F)zLight

transmittance (%)

Glass 1.13 75Glass double layer 0.65 70Fiberglass 1.00 75Corrugated polycarbonate 1.20 75Polyethylene 1.15 65Polyethylene double layer 0.70 60Polycarbonate biwall 0.65 60Polycarbonate triple-wall 0.58 56Acrylic biwall 0.65 60IR filmy 1.00 65IR film double layer 0.61 60Solid insulation foam 0.23 0Concrete block 0.51 0Poured concrete 0.75 0Insulated concrete 0.13 0z1 Btu = 1.0551 kJ, 1 ft2 = 0.0929 m2, (�F-32)/1.8 = �C.yInfrared-blocking polyethylene film.

Fig. 1. When a user begins using Virtual Grower software, this is the main designpage and opening screen view. The ‘‘Add New Greenhouse’’ button populates themiddle of the window with values for greenhouse name, length, width, knee wallheight, materials, fuel types, infiltration, and heating system efficiency. Users can thenchange these values in the drop-down windows, or describe the house in more detailthrough additional buttons on this screen. Additional areas of Virtual Grower areaccessible through the tabs at the top of the screen; 1 ft = 0.3048 m, 1 gal = 3.7854 L.

780 • August 2010 20(4)

TECHNOLOGY AND PRODUCT REPORTS

unit content in units most commonlyassociated with that fuel type. Theoriginal source for British thermal unitvalue (Bartok, 2005; Knaebe, 2004)assumed 70% to 80% combustion ef-ficiency, but do not state which oneswere assumed to be 70% or 80%.Because the user estimates heating

system efficiency independently, Vir-tual Grower corrects for the assump-tion of 70% to 80% efficiency bydividing the source material Britishthermal unit value by 0.75 (an aver-age or compromise between the 70%and 80% efficiencies) before it is usedto calculate heating costs.

When a fuel is selected, a cost forthat material per unit is loaded, butcan be changed by the user. As withgreenhouse materials, there is a fea-ture that allows a user to add a fuelwith unique energy contents andcosts, thus experimental or theoreti-cal materials can be tested easily.

Describing the heatingschedule

After clicking on the ‘‘HeatingSchedule’’ tab, a user can set temper-atures of the different greenhouses.Thereare fourpreprogrammed‘‘classes’’of schedule options: ‘‘No Heating,’’‘‘Its Day When The Sun Is Up,’’ ‘‘ItsDay When I Say It Is,’’ and ‘‘ConstantTemperatures.’’ The ‘‘No Heating’’setting simply does not heat that green-house for the time frame selected, andno cost is calculated, while the ‘‘Con-stant Temperature’’ setting sets a singletemperature for the period of the yearselected.

In the ‘‘Its Day When The Sun IsUp’’ setting, the daytime temperaturesetpoint is activated when the suncomes up, based on your location inthe country. When the sun is not up,the night temperature is maintained.In this selection, the thermoperiodmatches the natural photoperiod orlight period. The ‘‘Its Day When I SayIt Is’’ setting allows for the daytimetemperature to be set to a specific timeperiod in the day. This time perioddoes not change from one day to thenext, unless the user changes the set-tings for that day.

In addition to the preprog-rammed sections, a user can define a‘‘Custom Schedule,’’ programmablefor individual hours in the day. Thisallows for programming different dayand night temperature treatments(DIF), or temporary, short-term dropsin temperature just before sunrise(DIP) treatments for certain cropsor certain times of the year (Erwinet al., 1994). All schedules are savable,thereby allowing a user to have thatschedule for future greenhouse de-signs. The program allows for complex,realistic scheduling for entire green-houses, but currently does not permitsubdividing greenhouse sections withtemporary partitions as is a commonpractice in greenhouse production.

Calculating heating costsOnce a greenhouse is created

and a heating schedule is assigned to

Table 2. In the section for air infiltration within Virtual Grower software, a usercan describe their greenhouse air leakage based on size and abundance of gaps,characteristics of ventilation systems, and other construction designs. The valueof air exchanges that each category/item adds to the baseline air exchange rate of0.5 air exchanges per hour is shown.

Leakage sourceAdditional

selection criteriaAir exchangesadded (no.)

Large gaps None 0.0Few 0.3Several 0.6A lot 1.2

Small gaps None 0.0Few 0.15Several 0.3A lot 0.6

Outside air intakes 0.75Gaps between greenhouse and vents 0.25Malfunctioning vent covers 0.5Office, headhouse, retail area,

and shipping areaz

Connected, but insulated area 0.25

Connected and not insulated 0.5zBaseline, default value is to assume this is a separate building.

Fig. 2. Advanced design options window in Virtual Grower, version 2.5. From thisscreen, a user can describe their greenhouse design and materials in detail, tailoringeach wall to match their existing or planned structure. The diagram of thegreenhouse at the bottom changes depending on the overall style of greenhouse thatis selected; 1 ft = 0.3048 m.

• August 2010 20(4) 781

that greenhouse, costs can be calcu-lated. To do this, heat gained by solarradiation (if daytime) is added to heatlost through conduction and convec-tion (if temperature outside is lowerthan the set point) for each hour ofthe day. The equations used are fromAldrich and Bartok (1994) and aredescribed in detail therein and brieflybelow.

For an estimate of heat gained bysolar radiation, the light energy perarea (from weather database) is mul-tiplied by the total footprint of thegreenhouse. To account for reflectionand evapotranspiration, it is assumedthat one-third is reflected from thegreenhouse materials and half of theremainder is used for evaporatingwater from plants and substrate sur-faces with that evaporated water leav-ing the greenhouse (assumption usedin Aldrich and Bartok, 1994). Thismeans that only 33% of the sun’senergy heats the greenhouse.

To calculate conduction over anygreenhouse surface, four values are

used: the temperatures on either sideof the surface (usually the tempera-ture set point of the greenhouse andthe current temperature outside), thearea of the surface, and a constantknown as the unit area thermal con-ductance value (U-value) of the sur-face material. By multiplying thesevalues together, the amount of heatper hour that is lost to that surface canbe found. This value can be multipliedby a correction factor based on windspeed, reported by Bartok (2005), toaccount for the effects of wind.

To calculate convection for anyparticular greenhouse, four values areused: the temperatures on the insideand outside of the greenhouse, thevolume of air within the greenhouse,and a correction factor representingthe relative leakiness of the green-house structure. By multiplying thedifference between the two tempera-tures by the volume and the correc-tion factor, an estimate of heat losscan be obtained. As with conduction,an additional correction factor to

account for the effects of wind canbe applied.

For the entire structure, conduc-tion is calculated for each wall, knee-wall, and the roof, while convectionis calculated for the whole structure.Adding the results together with solargain provides the net energy lost ata given temperature. If the result isnegative (i.e., the greenhouse is cool-ing off), the number of British ther-mal units lost is assumed to be thenumber of British thermal units addedto the greenhouse by its heaters, tokeep it at its current temperature.

Cost is then calculated by divid-ing British thermal units lost by theamount of heat gained from burningone unit of selected fuel, adjusted forthe given heater and combustion ef-ficiencies. This gives us the amount offuel needed to keep the greenhouseup to temperature. Finally, the costto heat the structure is calculated bymultiplying the units of fuel burnedby the cost per unit specified for thatgreenhouse. Cost is expressed per

Table 3. Heating system efficiency is estimated in Virtual Grower software by a matrix of ventilation type, heat deliverymethod, and maintenance frequency. Efficiency values are shown for each possible selection within the matrix. A user selectsone term from each section (‘‘Ventilation,’’ ‘‘Heat delivery,’’ and ‘‘Maintenance’’), the three values are multiplied, and a finalsystem efficiency is estimated. Note that in ‘‘Heat delivery,’’ there are three sections for delivery method, but only one type isused from forced air, exposed pipes, or in floor.

Ventilation type, heating system type, and resulting efficiency (%)

Heater type Gravity Power or fan-assisted Condensing Separate

Standard unit heater 65 78 — 80High-efficiency unit heater 70 80 — 82Hot water boiler 68 82 93 84Steam boiler 65 78 90 80

Heat delivery method, location, and resulting efficiency (%)Forced air

Heater type Above bench Below bench Inflatable tubesabove bench

Inflatable tubesbelow bench

Standard unit heater 60 65 65 70High-efficiency unit heater 65 70 75 80

-Or-Exposed pipes

Perimeter Below bench Finned perimeter Finned below benchHot water pipes 85 88 87 90Steam pipes 80 83 82 85

-Or-In floor

Not insulated InsulatedHeated floors 80 90

Maintenance frequency, age of heating system, and resulting efficiency (%)Age Weekly or monthly Annually Every few years NeverNew (1–2 years) 100 99 98 97Intermediate (3–5 years) 99 96 94 92Old (over 6 years) 98 94 92 90

782 • August 2010 20(4)

TECHNOLOGY AND PRODUCT REPORTS

year and per month on a square footand total basis.

An alternative method of calcu-lating a building’s heat requirementsis the use of a ‘‘Heating Degree Day’’model. This method bases heat needson the average temperature duringa day. If the average temperature atthe specific location is below 65 �F,heating is required. A list of heatingdegree days for different locations canbe found online (Cornell University,2010). The heating degree days thatVirtual Grower predicts and this da-tabase does are in close agreement(typically less than 1% variation). Onsome days, however, the heating de-gree day model predicts no heat isneeded when Virtual Grower wouldpredict some heat is needed. For ex-ample, assume your greenhouse hasa set point of 65 �F. If a location

(hypothetical) has an average temper-ature of 65 �F, has a 12-h day of 75 �F,and a nighttime of 55 �F, VirtualGrower would heat the house for 12 h(during the night) and the heating de-gree day model would predict that noheat is needed.

GENERATING A REPORT. Fromthe ‘‘Costs’’ page, a text-based reportcan be generated, copied, and savedin any word-processing program. Thereport describes the current settings(greenhouse design considerations,fuel types, etc.) and costs associatedwith heating that greenhouse.

REAL-TIME DATA. For long sim-ulation periods, historical data hasvalue, but for immediate, short-termpredictions and validation, the ‘‘RealTime Data’’ feature can allow a user tocontrol costs for heating over the next2 d, provided the user has an active

internet connection. In this section,there is a box into which a zip codecan be typed. Once entered, pressingthe ‘‘Calculate Heating Costs’’ but-ton retrieves the 2-d forecast froma website (National Oceanic and At-mospheric Administration, 2010)that is specific to that zip code. In afew seconds, the forecast appears inthe top part of the window, brokendown in 3-h increments, and the costsfor each selected greenhouse and itsBritish thermal unit usage will appearin the bottom box. The heating costsprediction depends on the greenhousessimulated and heating schedule. Thisfeature is only functional for U.S. sites,except Alaska.

The ‘‘Calculate Heating Costs’’button is inactivated for a few minutesafter each web service call to the webserver. This will cut down on usersinadvertently making multiple datarequests and save on bandwidth.

At any time during simulation or‘‘constructing’’ a greenhouse, the usercan save the greenhouse setting fromthe ‘‘Main’’ page by clicking ‘‘File-SaveSettings’’ and naming the greenhousedata file (.gdf). This enables a user toreload the greenhouse in other VirtualGrower sessions. Exiting out of theprogram at any point will also promptthe user to save the settings.

Virtual Grower allows the choiceof U.S. or metric units. From any page,a user can choose U.S. or metric; de-fault settings are in U.S. units. Makingthe change will convert the units to thatconvention throughout the program.

MODEL USE AND VALIDATION.Extensive field validation in commer-cial greenhouses has been done insome locations of the United States.In the midwestern U.S., the Centerfor Innovative Food Technology(Toledo, OH) conducts energy assess-ments at greenhouse operations andtheir engineers have used VirtualGrower to analyze energy consump-tion for winter heating at several dif-ferent sites in Ohio. These evaluationshave included single-span and gutter-connected multiple-span structuresconstructed of glass, polyethylene, pol-ycarbonate, and fiberglass, and thoseoperated under full season and partialseason schedules. Four examples offield validation are presented as casestudies below.

C A S E S T U D Y 1 : H E A T E R

REPLACEMENT. A 2.5-acre greenhouseoperation in Delta, OH, has heating

Table 4. List of fuel types, energy content, and default units in the database usedby Virtual Grower. Users can select from the different fuel types and assigndifferent fuels to different greenhouses to simulate fuel consumption based onother user-defined selections. All values were obtained from Bartok (2005) andKnaebe (2004).

Fuel type Energy (Btu)zUnit for

energetic amountz

Fuel oil #2 138,500 galFuel oil #4 145,000 galFuel oil #6 153,000 galNatural gas 100,000 thermPropane 92,500 galHard coal 26,000,000 tonSoft coal 24,000,000 tonElectricity 3,412 kWhMethane 1,000 ft3

Methanol 57,000 galLandfill gas 500 ft3

Butane 130,000 galEthanol 76,000 galKerosene 135,000 galWaste oil 125,000 galBiodiesel 120,000 galGasoline 125,000 galSoft wood 12,500,000 cordHard wood 21,000,000 cordHogged wood 18,000,000 tonSawdust (green, 45% moisture) 9,000,000 tonSawdust (dry) 16,000,000 tonWood chips 7,600,000 tonBark 9,750,000 tonWood pellets (10% moisture) 16,000,000 tonRubber pellets 33,000,000 tonPlastic 19,000 lbCorn shells 16,000,000 tonCorn cobs 16,500,000 tonSwitchgrass 11,625,000 tonz1 Btu = 1.0551 kJ, 1 gal = 3.7854 L, 1 therm = 105.4804 MJ, 1 ton = 0.9072 Mg, 1 ft3 = 0.0283 m3, 1 cord =3.6246 m3, 1 lb = 0.4536 kg.

• August 2010 20(4) 783

needs from October to the end ofMarch. The portion of the green-house slated for upgrading consistsof a combination of single-span andgutter-connected houses, all withdouble-layer polyethylene glazing onthe walls and roof. The facility usespropane, which was assumed to be$1.50 per gal, an approximate 3-yearaverage for this facility. Baseline con-ditions were modeled at a day/nighttemperature of 60/55 �F, with over-all heating system efficiency of 60%due to older heaters, gravity-ventedventilation, and good heat distributionsystem from inflated poly-tubes un-der the bench. Virtual Grower esti-mated a cost of $34,000 for propanein a typical year (Table 5). A 3-yearaverage cost using the facility’s fuelbills yielded an estimate of $32,200cost for fuel or a difference of 5.6%.

A heating system upgrade wassimulated by the changing heatingsystem efficiency from 60% to 80%.Virtual Grower estimated a savings of$8520 per year. After heater installa-tion, separate engineering estimatesand measurements of fuel consump-tion for the initial year predicted anannual savings of around $8100 ora difference of 5.1%.

CASE STUDY 2: SIDE-WALL

CURTAINS. A 0.5-acre gutter-connectedfacility in Bowling Green, OH, was in-vestigating the practicality of savingenergy by inexpensive temporary sidecurtains. The facility has a combinationof twin-wall polycarbonate and double-polyethylene roof, but mostly single-layered glass walls. The heating seasonin this greenhouse is from December toMarch, and uses natural gas-fired unitheaters. Baseline heating costs at day/

night temperature settings of 65/60 �Fwere estimated to be $5700 assuming$7 per million Btu, a 3-year average gascost (Table 5). Actual gas costs over theprevious 3 years were $6120 or a differ-ence of 6.8%.

A change in the model fromsingle layer glass walls to double layer(glass or different plastic materials)predicted possible savings of $300per year, with little variation dueto the choice of material to be tem-porarily hung on the glass walls. Withthis information, the operation in-stalled a plastic sheet on the west glasswall to enhance insulation throughthis winter operation. The installationwas simple and low-cost and thecurtain is manually lifted and lowered.Cost savings estimated from thegrower confirms that the fuel savingsfrom this installation are in the rangeof $300 to $400, or a difference ofabout 12.5% (assuming $350 actualsavings).

C A S E S T U D Y 3 : H E A T E R

REPLACEMENT. Twenty old gravity-vented unit heaters in Swanton, OH,were in need of replacement withhigh-efficiency power-vented models.The heaters were located in two gut-ter-connected ranges with double-layer polyethylene glazing, and wereoperated under varying heating sched-ules through winter and spring usingnatural gas-fired unit heaters. The tworanges were modeled at 60% heatingefficiency and were assumed to cost$9.50 per million Btu, a 2-year averagecost for that facility. The model pre-dicted those ranges to have a cost ofaround $61,800, which correspondedwell to the previous years and engi-neering estimates for consumption

breakdown for the houses, estimatedat around $59,280 (a difference of4.4%; Table 5). The model was re-run with heating efficiencies of 80%to reflect the upgraded heaters, andshowed a savings of around $15,450per year. Separate engineering esti-mates also confirmed annual savingsof about $15,500 per year with theheater replacement, or a difference of–0.4%.

CASE STUDY 4: BOILER SIZING

AND SELECTION. A new 5-acre com-mercial greenhouse facility was beingplanned in Avon, OH. The glass-glazed gutter-connected facility in-corporated the latest energy-efficientfeatures includingzonecontrol, thermaland shade curtains, efficient hydronicwarm-floor and overhead heating sys-tem, and efficient irrigation and materialhandling systems. Energy modelingwith Virtual Grower was performed toestimate the benefits of the renewablefuel heating system, wood chip biomassover conventional natural gas-firedsystem. Virtual Grower assumed wastebiomass (locally sourced waste sawdust, bark, and chips) at $25 per ton,and used individual bays (part of thegutter connected houses) for Septem-ber to December. The model pre-dicted a cost of heating to be about$56,000 per year compared with theactual operational cost for the biomassheating system over the last two heat-ing seasons of about $50,000 per year,a difference of 11.9% (Table 5).

In general, these case studiesillustrate that when full seasons oryears are simulated, predictions havebeen found to be within about 12%of actual costs with greenhouses main-tained at fairly consistent temperatures

Table 5. Virtual Grower was used to simulate baseline conditions in commercial greenhouses and those predicted costs werecompared with actual heating bills. Following baseline description, Virtual Grower was used to simulate a change inoperating conditions and evaluate for financial savings. After implementation, heating bills were compared with predictedvalues for assessment of differences. N/A = not applicable in this case due to new construction. All facilities used heating from1 Sept. to 31 May.

Case study

Baseline conditions Change or savings

Model ($) Actual ($) Difference (%) Model ($) Actual ($) Difference (%)

Heater replacementz 34,000 32,200 5.6 8,520 8,100 5.1Side curtainsy 5,700 6,120 –6.8 300 350 –12.5Heater replacementx 61,800 59,280 4.4 15,450 15,500 –0.3Boiler selectionw N/A N/A N/A 55,900 50,000 11.9zModel assumptions: each house 180 · 30 ft (54.9 · 9.1 m), single span, double polyethylene sides and walls, day/night 60/55 �F (15.6/12.8 �C), propane at $1.50 per gal($0.40/L), 60% heating efficiency (baseline), 80% heating efficiency (savings), Toledo, OH, weather database.yModel assumptions: four-bay gutter connected house, each bay 120 · 27 ft (36.6 · 8.2 m), glass sidewalls (baseline), double layer walls (savings), double polyethylene roof,day/night 65/60 �F (22.0/15.6 �C), natural gas at $0.70 per therm ($0.0066/MJ), Toledo, OH, weather database.xModel assumptions: gutter connected houses [four bay and five bay, each approx 3200 ft2 (297.3 m2)], double layer polyethylene, day/night 60/55 �F, natural gas at $0.95per therm ($0.0090/MJ), 60% heating efficiency (baseline), 80% efficiency (savings), Toledo, OH, weather database.wModel assumptions: gutter connected glass house with 14 bays, each bay 432 · 40 ft (131.7 · 12.2 m), 65 �F constant temperature, wood chips at $25 per ton ($27.55/Mg),80% heating efficiency, Cleveland, OH, weather database, 25% space utilization in September and October, 50% space utilization in November and December.

784 • August 2010 20(4)

TECHNOLOGY AND PRODUCT REPORTS

and heating schedules. Additionally, arange of different project scopes can besimulated, from new construction tosmaller greenhouse modifications,with adequate results. Our experienceshows that when shorter time framesare simulated, variations up to 50%have resulted at sites where green-house space is occupied and heatedprogressively with growing schedulesthrough the season relying on thehistorical databases (2 weeks to 1month). This may be due to seasonalor short-term deviations from the‘‘typical’’ historical weather data usedin Virtual Grower. The historical per-spective of the climate and solar data-bases for different sites used by VirtualGrower and the resulting model pre-dictions are therefore most useful forcomparative analysis of efficiency im-provement features over multiple sea-sons and to derive baseline operationalcharacterizations.

Future software developmentVirtual Grower continues to be

developed. Adding plant growth anddevelopment models will allow forscheduling and an assessment of plantquality. Facing increasing fuel costs,growers are always looking for ways toreduce their heating bills, sometimesto the detriment of crops. In fact,intuition that tells growers to lowera setpoint to curb energy use maymake the problem worse. Runkleet al. (2009) have calculated that insome crops and for some markets,more cost-effective and energy-efficientcrops can be grown in warmer green-houses than cooler houses due to delaysin development. That is, slower de-velopment in cooler greenhouses leadsto longer production times, so eventhough the per day energy consump-tion is lower, the total energy con-sumption is higher to finish the crop.

Improving the realism in heatingsystems and partitioning of green-houses would provide more realisticsimulation opportunities. For exam-ple, early in growing seasons, growershang temporary plastic ‘‘curtains’’ be-tween greenhouse sections and heatonly parts of large ranges. These canbe simulated in the current softwareonly through complex, non-intuitive‘‘work-arounds,’’ so adding this asa design or scheduling option wouldbe helpful.

Increasingly, growers have mul-tistage heating systems with different

heaters and potential fuel sources toprovide baseline or emergency heat.The current version only applies a sin-gle heating system and fuel type fora greenhouse, so a multitiered heatingsystem option would add more flexi-bility and realism for simulations, andhelp to identify the investment valueof these systems.

Carbon footprints could be cal-culated from the existing software’sframework, and predictions of plantpest outbreaks and water use couldalso be folded in, with linkages to thehistorical weather database alreadyused. There are advantages to keepingthese models separate (simplicity, forexample), but a significant advantageof observing interactions and poten-tial for determining the impacts ofoptimizing for one feature (plantgrowth, for example) that may occuron other features (energy consump-tion and likelihood of pest outbreaks)would be possible.

Setup requirementsVirtual Grower 2.5 was written in

Visual Basic.NET (Microsoft, Red-mond, WA) architecture and uses Ac-cess (Microsoft) database files. It is onlycompatible with computers runningWindows 98, ME, 2000, XP, Vista(Microsoft), or later. It does not oper-ate on 64-bit processors, though cur-rent efforts on the program are makingthe software platform independent.User ‘‘help’’ is provided throughacom-prehensive manual through the pro-gram as well as with an e-mail address(USDA-ARS@utoledo.edu). The latestversions of the software can be down-loaded for free from the internet (U.S.Department of Agriculture, 2006).

Literature citedAldrich, R.A. and J.W. Bartok, Jr. 1994.Greenhouse environment, p. 61–72. In:M. Sailus, C. Napierala, and M. Sanders(eds.). Greenhouse engineering. NaturalResource Agr. Eng. Serv. Publ. 33.

Bartok, J.W., Jr. 2005. Greenhouse man-agement/engineering: Fuels and alter-nate heat sources for commercialgreenhouses. 27 Nov. 2009. <http://www.umass.edu/umext/floriculture/fact_sheets/greenhouse_management/jb_fuels.htm>.

Brumfield, R.G. 1992. Greenhouse costaccounting: A computer program formaking management decisions. Hort-Technology 2:420–424.

Cornell University. 2010. Comparativeclimate data: Normal heating degreedays – base 65 degrees F. 4 May 2010.<http://www.nrcc.cornell.edu/ccd/nrmhdd.html>.

Erwin, J., P. Velguth, and R. Heins. 1994.Day/night temperature environment af-fects cell elongation but not division inLilium longiflorum Thunb. J. Expt. Bot.45:1019–1025.

Fisher, P. and C. Donnelly. 2002. De-velopment of a financial model to evaluateinvestment in supplemental lighting forgreenhouse floriculture production. ActaHort. 580:191–196.

Giacomelli, G.A. and W.J. Roberts. 1993.Greenhouse covering systems. HortTech-nology 3:50–57.

Hochmuth, R.C. and G.J. Hochmuth.1993. Use of plastic in greenhouse vege-table production in the United States.HortTechnology 3:20–27.

Knaebe, M. 2004. Fuel value and powercalculators. Techline Nwsl., WOE-3. U.S.Dept. Agr., Forest Products Lab, Madi-son, WI.

Marion, W. and K. Urban. 1995. User’smanual for TMY2s: Typical meteorolog-ical years. Natl. Renewable Energy Lab.Special Publ. 463-7668.

Moccaldi, L.A. and E.S. Runkle. 2007.Modeling the effects of temperature andphotosynthetic daily light integral ongrowth and flowering of Salvia splendensand Tagetes patula. J. Amer. Soc. Hort.Sci. 132:283–288.

National Oceanic and Atmospheric Ad-ministration. 2010. National weather ser-v i c e . 4 M a y 2 0 1 0 . <h t t p : / /forecast.weather.gov/(unique-code-for-location)>.

Runkle, E., J. Frantz, and M. Blanchard.2009. Energy efficient annuals: Schedul-ing bedding plants. Greenhouse Grower27(4):41–44.

Simpkins, J.C., D.R. Mears, W.J. Roberts,and H.W. Janes. 1984. Evaluation of anexperimental greenhouse film with im-proved energy performance. ASAE Paper84-4033. Amer. Soc. Agr. Eng., St.Joseph, MI.

U.S. Department of Agriculture. 2006.Virtual Grower. 4 May 2010. <http://www.virtualgrower.net>.

U.S. Department of Energy. 2009.Monthly energy review. 27 Nov. 2009.<http://www.eia.doe.gov/emeu/mer/prices.html>.

Wheeler, R.M. 1992. Gas-exchange mea-surements using a large, closed plant growthchamber. HortScience 27:777–780.

• August 2010 20(4) 785