ABSTRACTIn the United States, intercity long-haul trucks idleapproximately 1,800 hrs per year primarily for sleeper cabhotel loads, consuming 838 million gallons of diesel fuel [1].The U.S. Department of Energy's National RenewableEnergy Laboratory (NREL) is working on solutions to thischallenge through the CoolCab project. The objective of theCoolCab project is to work closely with industry to designefficient thermal management systems for long-haul trucksthat keep the cab comfortable with minimized engine idling.Truck engine idling is primarily done to heat or cool the cab/sleeper, keep the fuel warm in cold weather, and keep theengine warm for cold temperature startup. Reducing thethermal load on the cab/sleeper will decrease air conditioningsystem requirements, improve efficiency, and help reducefuel use. To help assess and improve idle reduction solutions,the CoolCalc software tool was developed. CoolCalc is aneasy-to-use, simplified, physics-based heating, ventilatingand air conditioning (HVAC) load estimation tool thatrequires no meshing, has flexible geometry, excludesunnecessary detail, and is less time-intensive than moredetailed computer-aided engineering (CAE) modelingapproaches. It is intended for rapid trade-off studies,technology impact estimation, and preliminary HVAC sizingdesign. It also complements more detailed and expensiveCAE tools by exploring and identifying regions of interest inthe design space. CoolCalc is built on NREL's OpenStudioplatform and is a plug-in extension of Google's SketchUpsoftware. This paper describes the CoolCalc tool, providesoutdoor long-haul truck thermal testing results, showsvalidation using these test results, and discusses futureapplications of the tool.

INTRODUCTIONHeating and air conditioning are two of the primary reasonsfor long-haul truck main engine operation when the vehicle isparked. In the United States, trucks that travel more than 500miles per day use 838 million gallons of fuel annually forovernight idling [1]. Including workday idling, over 2 billiongallons of fuel are used annually for truck idling [2]. Byreducing thermal loads and improving efficiency, there is agreat opportunity to reduce the fuel used and emissionscreated by idling. Reducing the thermal load for truck cab/sleepers will enable cost-effective idle reduction solutions. Ifthe fuel savings from new technologies can provide a 3- to 5-year payback time, fleet owners will be economicallymotivated to incorporate them. This provides a pathway torapid adoption of effective thermal load reduction solutions.

The U.S. Department of Energy's (DOE's) NationalRenewable Energy Laboratory (NREL) CoolCab project isresearching efficient thermal management systems that keepthe cab occupants comfortable without the need for engineidling. The CoolCab research approach is to reduce thermalloads, concentrate on occupant thermal comfort, andmaximize equipment efficiency. By working with industrypartners to develop and apply commercially viable solutionsthat reduce idling fuel use, both national energy security andsustainability will be improved. To achieve this goal, NRELis developing tools and test methods to assess idle reductiontechnologies. The truck cab industry needs a high-levelanalysis tool to predict thermal loads that are used to evaluateload reduction technologies and their impact on climate-control fuel use.

To meet this need, NREL has developed CoolCalc, asoftware tool to assist the industry in reducing climate controlloads for heavy-duty vehicles. CoolCalc is a heating,ventilating and air conditioning (HVAC) load estimation tool

CoolCalc: A Long-Haul Truck Thermal LoadEstimation Tool

2011-01-0656Published

04/12/2011

Jason A. Lustbader, John P. Rugh, Brianna R. Rister and Travis S. VensonNational Renewable Energy Laboratory

doi:10.4271/2011-01-0656

NREL/CP-5400-52452. Posted with permission. Presented at the SAE 2011 World Congress & Exhibition.

that enables rapid exploration of idle reduction design optionsfor a range of climates.

APPROACHCOOLCALC MODEL DEVELOPMENTCoolCalc is an easy-to-use, simplified, physics-based HVACload estimation software tool that requires no meshing, hasflexible geometry, excludes unnecessary detail, and is lesstime-intensive than more detailed computer-aidedengineering (CAE) modeling approaches. It is intended forrapid trade-off studies, technology impact estimation, andpreliminary HVAC sizing design. It complements moredetailed and expensive CAE tools.

CoolCalc is built on NREL's OpenStudio platform. This wasdone to accelerate development and leverage previous andongoing DOE investments. OpenStudio was developed atNREL and released in 2008. It is a plug-in extension ofGoogle's SketchUp software. DOE's EnergyPlus is used asthe heat transfer solver. EnergyPlus is a DOE-fundedsoftware package designed for building efficiency analysis,which was found to be general enough to extend to truck cabthermal modeling.

Unlike previous building thermal simulation programs,EnergyPlus provides a fully integrated simulation where thebuilding HVAC zones, system, and plant (source) are solvedtogether. This provides a more physically realistic solutionand allows for more detailed control implementation. Heattransfer is described by a set of time-dependent energy andmoisture balances, and the resulting ordinary differentialequations are solved using a predictor-corrector approach.For solar loading, an anisotropic radiance model is used,allowing the superposition of three components: isotropicradiance, point source circumsolar brightening at the sun, andhorizon brightening. The sun position is tracked as a functionof geographic location, time of year, and time of day. Ashading model is implemented that accounts for shadowing ofsurfaces by other surfaces. EnergyPlus's window model isbased on Lawrence Berkeley National Laboratory'sWINDOW program algorithms and uses solar transmittance,reflectance, and absorptance properties. Full spectral analysisis also possible. All surfaces are treated as gray bodies. Theexternal surface heat transfer balance includes shortwavesolar radiation, longwave thermal radiation, convection, andconduction through the walls. Conduction is one dimensionalthrough the thickness of a wall. Walls are defined using aseries stack-up of materials that allow for thermal storage.The interior surface heat transfer modes include shortwavesolar radiation from windows, longwave radiation, andconvection. The internal radiation view factors areapproximated by a ratio of “seen” areas, then corrected forreciprocity and completeness. A detailed description ofEnergyPlus's modeling and solution methods is beyond the

scope of this paper; for further details see the extensivedocumentation available [3].

While CoolCalc is flexible and does not dictate a specificprocess, a typical workflow (illustrated in Figure 1) mightbegin with the creation of geometry using the Parametric Cabcreation tool (Figure 2). The Parametric Cab creation windowhas a series of tabs across the top, one for each air zone in themodel. Each tab has a list of available parametric variablesthat will modify the geometry. These variables and theparametric geometry relationships are determined by a modeldefinition file created in the geometry coding framework. Theuser can also switch between the available parametric cabmodels to find one that that best suits their needs. To the rightof the units, the allowable variable range is displayed. Toillustrate this parametric capability, the windscreen angle waschanged from 60° to 80° (Figure 3). The cab model quicklyupdates, allowing for fast modification of the geometry.

Figure 1. Typical CoolCalc workflow

Figure 2. Parametric Cab geometry creation

Figure 3. Parametric Cab with modified windscreenangle

Once the cab geometry is established using the ParametricCab tool, it can be manually modified by the user. Figure 4shows an example of a user adding an additional sidelight tothe sleeper cab. The sleeper cab sidewall was activated bydouble clicking on the surface. Once activated, the SketchUpdrawing tools can be used to modify the geometry. Thedashed lines are construction lines that were created to helpquickly draw the sidelight and can be easily hidden or deletedlater. The pencil tool was then used to trace out theconstruction lines. The pencil tooltip icon can be seen in thetop left corner of the sidelight. Once the window shape isclosed by the pencil tool, it is automatically recognized as awindow and assigned default properties.

Figure 4. Manual modification of geometry: adding asidelight

In EnergyPlus, every component of the model, e.g., walls,materials, location, and solver time step, is treated as anobject. In CoolCalc, to modify or define new objects theObject Browser tool is opened (Figure 5). On the left side ofthe Object Browser window is the object tree, which shows

all the objects that are available in the model and allowscreation of new objects. Below the object tree is the librarywindow. The library window allows the user to load andmanipulate additional libraries of objects. These objects canthen be added to the current model by dragging and droppingthem into the object tree. Below the specific object windowinterface, to the right of the object tree, is the text editingwindow. This window allows for manual modification of thecurrent object, giving full control to advanced users. Forobjects where no specific interface has been developed, thetext editing window will comprise the entire right side of thesplit window. All object windows also provide a “Comment”option in the upper right corner to help users document theirobject assumptions.

To modify or define new materials, a material object isselected in the object tree (Figure 5). Based on this object treeselection, the material definition window is displayed to theright of the object tree. The material definition windowprovides text boxes or pulldown menus for all the basicmaterial thermal properties: name, roughness, thickness,conductivity, density, specific heat, and radiationabsorptance.

Figure 5. Object Browser and Material DefinitionWindow

Each surface in CoolCalc is treated as multiple layer, one-dimensional conduction, forming a “sandwich” typestructure. To define this layered structure, a constructionobject is used. Once again navigating the object tree, aconstruction object is selected. This changes the currentobject window (right side) to display the constructiondefinition window (Figure 6). In this window, the user selectsthe materials to include in the construction and can changetheir order. Materials assigned to the inner and outer layerswill determine the solar radiation properties and the texturedisplayed in the texture-rendering mode.

Figure 6. Construction Definition Window

Figure 7 shows the model in construction rendering mode. Inthis mode, the inner and outer surfaces are colored by theirrespective construction's material textures. The window to theleft of the cab is the Construction Palette. It allows sortingand selection of constructions and their application to themodel using a point-and-click paint can tooltip.

Figure 7. Construction palette window and texturerendering mode

Before solving the model, a weather file is selected. There arecurrently Typical Mean Year data available for 2,100locations worldwide. Custom weather data can also beentered. A simulation period of one day to one year is alsoselected. Once the model is solved, the results can bedisplayed within the interface. Figure 8 shows the exteriortemperatures displayed on the truck cab/sleeper.

Figure 8. Example temperature distribution

EXPERIMENTALTo validate the CoolCalc modeling approach, outdoorthermal soak testing of a Kenworth Truck Company T660Class 8 sleeper cab truck was conducted at NREL's VehicleThermal Soak Test Facility (Figure 9). The truck was parkedon level ground at an elevation of 5,853 feet, 39.7 degreeslatitude, −105.2 degrees longitude, facing due south. The testvehicle was instrumented with 14 exterior and 28 interior k-type thermocouples. The thermocouples were calibratedusing a Hart Scientific 7103 Micro-Bath for an overalluncertainty of less than ±0.42°C. The thermocouples werestrategically positioned to accurately characterize the vehiclethermal distribution and collect validation data. Surfacethermocouples were attached using an Omega thermallyconductive epoxy. Radiation shields were used for airthermocouples to minimize errors due to direct solarradiation. A pyranometer was placed on the instrument panelof each vehicle to confirm sunrise times, and humiditymeasurements were taken. An IOTech LogBook 360 dataacquisition system was used and located in the truck'stoolbox. Data was collected every second and reduced to oneminute averages over 24-hour periods. Solar radiation, windspeed, cloud coverage, relative humidity, and wind directionwere also measured at NREL's Solar Radiation ResearchLaboratory, which is located near the test site.

Thermal soak tests were conducted from August throughNovember 2009. The truck was exposed to daytime solarloads in an engine-off condition. An air exchange rate testwas also conducted to help characterize thermal behavior. Forthis measurement, a small amount of sulfur hexafluoride(SF6) was injected into the cab, and the concentration decay

was monitored. Samples were taken every 5 minutes for 3hours, and the air exchange rate was calculated from the SF6concentration decay rate.

Figure 9. Test trucks at NREL's Vehicle Thermal SoakTest Facility

VALIDATIONTo validate the CoolCalc simulation approach, a KenworthT660 thermal model was developed. When available,information from Kenworth or testing was used to define theCoolCalc model parameters; otherwise, engineering estimateswere made. Since CoolCalc does not use meshing, allsurfaces must be planar. Therefore, the computer-aideddesign model provided by Kenworth was simplified. Figure10 shows the model geometry compared to a photograph ofthe vehicle. The red lines indicated the domain boundary usedfor the model. This simplification process inherently requiressome approximation; however, an effort was made toaccurately represent the geometry.

Figure 10. CoolCalc validation geometry

The vehicle was modeled with four air zones: Cab, Sleeper,Fairing, and Toolbox. The full EnergyPlus exterior radiationcalculations were used; however, a simplified version of theinterior radiation calculations was applied to increasegeometry flexibility. The simple interior radiation modelassumes that all beam radiation that passes through thewindows falls on the floor of that zone and any reflectedradiation is uniformly distributed on all interior surfaces. A

previous case study was conducted that indicated that thisresults in increasing, but acceptable, differences duringwinter months due to the changing sun angle. The detailedsurface convection algorithms were used that account forboth natural and forced convection. The model time step wasset to 1 minute intervals.

The glass properties for the cab and sleeper were obtainedfrom Kenworth and its suppliers as were the materials andconstructions for the vehicle walls. The overall resistivity ofthe walls was reduced to account for impacts of structuralmembers and other likely disruptions to insulation in a realvehicle. The paint properties were not available and wereestimated based on previous experience with vehicle paints.

Four internal masses were defined to represent interiorobjects. The first represents the seats in the cab.Specifications on vehicle seats were used to estimate the size,weight, and material properties. The next two internal massesrepresent the dashboard and the sleeper closets. Both wereassumed to be polyurethane plastic, and the volume andsurface area of each were estimated using truck geometry.Lastly, the two beds were modeled. Information on these wassupplied by Kenworth. It was found that internal massassumptions can have a significant impact on thermal results,and further work on internal mass modeling approaches isplanned.

Since the chassis and hood would normally shade theunderlying surfaces and were excluded from the solutiondomain, the solar load was removed from both surfaces. Thefirewall boundary condition was set to represent naturalconvection at 20% above ambient temperature. The exteriorsurfaces of the vehicle floor were exposed to wind at ambienttemperature.

The measured air infiltration rate, as described in theexperimental section, was applied to the cab and sleeper.Since this model does not include a fluid flow solution, thecross-mixing between zones had to be estimated.Incorporating a simplified means for estimating this isanother possible improvement for the future.

RESULTSThe CoolCalc concept was validated by comparing theKenworth T660 thermal model results with experimentaltemperature and onsite weather data. Figure 11 shows theweather data used for the three consecutive validation days.The data set has been normalized at the request of industrypartners. This data set captures a range of conditions. Thefirst day was overcast, cooler, and had low wind. The secondtwo days were both clear, but had different wind andtemperature characteristics.

Figure 11. Normalized weather data for three validationdays

The model solution time for the three-day simulation with 1-minute intervals was 29.4 sec ±0.1 sec over three repeatedruns using a Dell Intel Core 2 Duo 3.16 GHz CPU with 3.25GB RAM desktop computer running the Windows XPoperating system. Figure 12 and Figure 13 show acomparison between the model and the measured airtemperatures. These graphs demonstrate good agreementbetween the model and the experimental data, both in thepeak soak temperature and the overall trends. Theexperimental averages for the cab and sleeper are an averageof six and eight distributed air temperatures, respectively.

Figure 12. Cab average air temperature comparison

Figure 13. Sleeper average air temperature comparison

Figure 14 summarizes the temperature difference between themodel and the test data for the time average (2-4 p.m.)sleeper and cab air temperatures. The maximum interior airtemperatures occurred during this time interval and thereforerepresent peak soak conditions. For the sunny days, thetemperature difference was less than 0.4°C, which is less thanthe measurement uncertainty. On the cloudy day, thetemperature difference between the data and the model was2°C. While the average peak soak temperature for day twocompares very closely with the experimental data, thedifferences with time can be seen in Figures 12 and 13.

Figure 14. Comparison of test data and model, timeaverage from 2-4 p.m.

The predicted exterior surface temperatures were alsocompared to the experimental results. Figure 15 shows theresults for the driver and passenger sleeper side wall surfacetemperatures. Because the truck is facing south, the driver-side surface temperatures rise in the morning, and then peakand decline as the sun passes over the vehicle. Likewise, the

passenger-side surfaces rise in the afternoon, peak, anddecline as the sun goes down. The higher frequency temporalvariability seen in the afternoon temperatures on day one forboth the experimental and the model results were caused bypassing clouds. On day three, a larger error is seen betweenthe model and the experimental data. Based on the high solarload and low wind speed, one possible cause for this could bethat the natural convection portion of the heat transfer isunderestimated in these conditions. However, furtherinvestigation would be needed to confirm this.

Figure 15. Sleeper driver and passenger exterior sidewalls

The concept validation results also show good agreementbetween the model and the experimental data for the othersurfaces that were compared. Figure 16 shows the insidesurface of the sleeper side windows. The model tends tooverestimate the temperature and respond faster. EnergyPlusdoes not currently have thermal mass for windows; therefore,this behavior is expected. This could also account for thehigher sensitivity to transients, such as passing clouds onDay1.

One of the larger differences between the model and theexperimental results is the windshield, shown in Figure 17and Figure 18. The temporal shift is more noticeable on thewindshield exterior. This might also be caused by the lack ofwindow thermal mass.

Figure 16. Comparison of sleeper interior side windows

Figure 17. Comparison of windshield exterior

Figure 18. Comparison of windshield interior

These validation results demonstrate that the CoolCalcmodeling tool can quickly and accurately estimate vehicletemperature distributions for air, surfaces, and glass. Whileinherent errors are expected when simplifying the geometry,this comparison to experimental results suggests thatsufficient accuracy is achieved for early design and rapidtrade-off analysis. Air conditioning tests were outside thescope of this first validation study. Matching the interiortemperatures well during a thermal soak provides someconfidence that the predicted air conditioning evaporator loadis realistic. Previous work on simulation of military-typevehicle geometry indicated that the evaporator thermal loadcalculations were reasonable. Future work will validate loadcalculations in more detail. By modifying exterior convectioncoefficients as a function of speed and adding engine heatloads, down the road HVAC loads could be investigated infuture work.

SUMMARY/CONCLUSIONSTo help develop solutions that reduce the 838 million gallonsof fuel used by long-haul trucks annually for overnight idlingin the United States, NREL's CoolCab project has developedan HVAC load estimation software tool called CoolCalc. Atypical workflow for modeling the thermal behavior of atractor trailer sleeper cab using CoolCalc was demonstrated.Vehicle geometry is first created using parametric andmanual tools. The user navigates and modifies solver objectsusing the Object Browser as needed to define modelparameters. Once the model is set up, the weather file isselected and the model is solved. Results can be displayedwithin the software environment or post-processed in detailusing output files.

Detailed experimental testing of a Kenworth T660 sleepercab truck was conducted at NREL to validate CoolCalc. Amodel of this vehicle was developed using informationprovided by Kenworth Truck Company, testing, andengineering assumptions. Comparison between the model andexperimental results collected over three days shows goodagreement both in trends and peak temperature values for avariety of weather conditions. The difference betweenexperimental and model peak soak air temperatures, anaverage from 2-4 p.m., was less than or equal to 0.4°C for thetwo sunny days and 2°C for the cloudy day. Surfacetemperature comparisons show that the effect of solarposition was captured accurately. Experimental testing for asecond validation case study has been completed, andmodeling is planned for the near future.

The ability of CoolCalc to quickly and accurately estimatevehicle thermal temperature distributions has beendemonstrated. Further validation with other vehicles andweather conditions is being conducted. Methods and tools arecurrently being developed to link CoolCalc thermal loadestimates to vehicle fuel use. The next step will be to apply

CoolCalc to study the impact of thermal load reductiontechnologies on idling and vehicle fuel use. This will fill animportant role in the CoolCab project's suite of experimentaland analytical tools that are being used to developcommercially viable idle reduction technologies incollaboration with industry partners.

REFERENCES1. Stodolsky, F., Gaines, L., Vyas, A. Analysis of TechnologyOptions to Reduce the Fuel Consumption of Idling Trucks.Argonne National Laboratory, ANL/ESD-43, June 2000.

2. Gaines, L., Vyas, A., Anderson, J. 2006, “Estimation ofFuel Use by Idling Commercial Trucks,” 85th AnnualMeeting of the Transportation Research Board, Washington,D.C., January 22-26, 2006, Paper No. 06-2567.

3. EnergyPlus Engineering Reference. http://appsl.eere.energy.gov/buildings/energyplus/pdfs/engineeringreference.pdf. October 11, 2010. Accessed on[10/25/2010]

CONTACT INFORMATIONJason A. LustbaderNational Renewable Energy LaboratoryJason.Lustbader@nrel.gov303-275-4443

John P. RughNational Renewable Energy LaboratoryJohn.Rugh@nrel.gov303-275-4413

ACKNOWLEDGMENTSThe authors would like to thank Lee Slezak and DavidAnderson, Technology Managers for DOE's AdvancedVehicle Technology Analysis and Evaluation for sponsoringthis work.

A special thanks to John Duffy and Stan DeLizo of KenworthTruck Company for their participation in and support of thiswork.

For their contributions to this project, thank you to thefollowing NREL employees: Rob Farrington, Charlie King,Mike Lammert, Ken Proc, Bob Rehn, and Brent Griffith.Additionally, thank you to Michael Birdsong and Peter Ellisfor your contributions to the program.

DEFINITIONS/ABBREVIATIONSCAE

computer-aided engineering

DOEU.S. Department of Energy

HVACheating, ventilation, and air conditioning

KenworthKenworth Truck Company

NRELNational Renewable Energy Laboratory

SF6sulfur hexafluoride

The Engineering Meetings Board has approved this paper for publication. It hassuccessfully completed SAE's peer review process under the supervision of the sessionorganizer. This process requires a minimum of three (3) reviews by industry experts.

ISSN 0148-7191

Positions and opinions advanced in this paper are those of the author(s) and notnecessarily those of SAE. The author is solely responsible for the content of the paper.

SAE Customer Service:Tel: 877-606-7323 (inside USA and Canada)Tel: 724-776-4970 (outside USA)Fax: 724-776-0790Email: CustomerService@sae.orgSAE Web Address: http://www.sae.orgPrinted in USA