S. De Backer, M.P. Diebold, and M.P. Milone, The Chemours Company
While there is an intuitive connection between wall brightness and theelectrical demand for illuminating a room, there has been little data reported to quantify this relationship. Here, we report the relative electrical loadneeded to light a room at the 500 lux level speciﬁed by European lighting standardEN12464-1:2011, as a function of wall brightness. A room was painted white, black,and two intermediate shades of gray, and room brightness was measured at multiplelocations and directions using four light levels (controlled with a light dimmer), forboth warm and cool ﬂuorescent bulbs. Results were compared to computer modeling.Based on these results, we determined the expected electrical requirements for roomspainted with over a dozen “colors of the year,” as designated by major décor coatingsmanufacturers. The results were compared to the electrical requirement for a whitewall. Signiﬁcant energy savings are possible when painting a dark wall white.
It is well known that reduction ofelectricity consumption in regions withhot summer climates can be realized bypainting the exterior surfaces of build-ings, particularly their roofs, white (oranother bright color). This maximizesthe reﬂection of solar radiation and sominimizes the amount of heat absorbedby these buildings. This concept is centu-ries old, but over the last decade it hasreceived renewed interest and attentionin the form of “white roof” or “whitebuilding” initiatives.1
While these beneﬁts are well recog-nized, there is a second, less recognized,energy advantage associated withbright colors. This savings applies to theamount of electricity required to lighta room brightly enough to carry out thetasks intended for that room. Dependingon the location and orientation of thetask (for example, writing on a table),and the locations of the light ﬁxtures,a signiﬁcant fraction of the incidentbrightness can be from light reﬂected bythe room walls. Obviously, the bright-ness of the walls will aﬀect the quantityof light reﬂected from it.
Electricity consumption for lightingis signiﬁcant. In the U.S. residentialsector, ineﬃcient incandescent lights arebeing actively replaced with more eﬃ-cient alternatives (primarily LEDs andcompact ﬂuorescent bulbs). However,this conversion is far from complete. Atthe end of 2017, more than half of thebulbs used in residential lighting werestill incandescent.2That year, 9% of totalresidential energy consumption was forlighting.3
Although Europe is overall more com-plete in their conversion from incandes-cent bulbs to more energy eﬃcient lightsources, a signiﬁcant amount of electric-ity is still consumed there for residentiallighting. In 2017, 13% of electricity usedin the UK was for all lighting.4Energyconsumption for lighting in ContinentalEurope is more diﬃcult to quantifybecause lighting is combined with elec-trical appliances in EU energy use sta-tistics. Globally, however, it is estimatedthat all lighting accounts for 15% of totalenergy consumption, and that lightingdemand will increase 50% between 2015and 2030.5
Painting a room with a bright colorresults in both direct and indirectelectrical savings. The direct savingsare obvious—as brightness increases,less light is needed. Indirect savingsare obtained during periods when theroom is actively cooled. The ultimatefate of most of the light generated in alight ﬁxture is heat. This is especiallytrue for incandescent lamps, where aslittle as 2% to 3% of the incoming energyis converted into visible light; the restis released as heat. By decreasing theenergy used to generate adequate light,we decrease the electrical load on thecooling system.
We can compare the electricity usedfor lighting to that used to cool build-ings. In the United States, it is estimatedthat 9.6% of the electricity used residen-tially in 2017 was for cooling.6Globally,in 2017, 10% of overall electrical usagewas for cooling.7While both types ofenergy consumption are roughly equal,we note that cool rooﬁng is a topic ofcurrent interest to the public, whilebright rooms is not.
That said, architects are aware thatreﬂectivity aﬀects room brightness,and during building design sometimesinclude wall reﬂectivity in their calcu-lations to determine the number andlocation of lighting ﬁxtures. The basisof these models is theoretical. Thesemodels have been used to optimizelighting in oﬃce settings,8but to ourknowledge, there has been no experi-mental data reported that measures theenergy required to adequately illumi-nate an actual room as a function of wallbrightness.
In this article, we detail the resultsof our experiment to determine electricenergy requirements as a function ofwall brightness for an average sizeinterior oﬃce room. We then comparethe results of our experiment to resultscalculated by a well-known architecturesoftware model.
An interior oﬃce was painted withachromatic paints of four brightnesses:brightest white, two shades of gray, andblack. The tristimulus Y reﬂectancevalues for these paints are 91, 60, 20, and5, respectively. A schematic of the room,showing the measurement and light ﬁx-ture locations, and a picture of the room,painted black, are shown inFigures 1and2. The room had a standard acousti-cal tile ceiling (tristimulus Y reﬂectancevalue of 84) and mid/dark carpeting(tristimulus Y reﬂectance value of 35).The ceiling height was 3.0 m, and mostmeasurements were made 90 cm fromthe ﬂoor. The exceptions were measure-ments made 60 cm underneath eachlighting ﬁxture. These measurementswere used to determine the luminositiesof the light sources, with the intentionbeing that all light measured at thosepositions comes directly from the lightsources, rather than being reﬂected fromthe walls.
Luminance was measured at sixlocations in the room, with the meteroriented at diﬀerent angles for theselocations (e.g., Horizontal, 45°, facingthe nearest wall or facing the interior)and with two diﬀerent dimmable ﬂu-orescent bulbs (25 W 4100K and 32 W6500K). There were two light ﬁxturesin the room, centrally located, andeach ﬁxture accommodated four bulbs.A diﬀusion panel was placed on eachﬁxture. Within a set wall brightnessand bulb type, four electrical powerlevels were explored at each location andorientation. Brightness was varied witha standard light dimmer.
A total of 776 brightness values weremeasured, including one set of mea-surements done in triplicate. Thesewere of the lighter gray room using the4100K lamps and measuring at all roomlocations and orientations. From this wecalculate an average standard deviationof 14.1 lux for light intensity and 0.5Watts for the power setting.
Typical results are shown inFigure 3.Here we display luminance values (onthe y-axis), for each room color (separateline), for the middle of the room with anup-facing orientation of the light meter,as a function of the four diﬀerent dimmersettings (on the x-axis) for the 4100Klamps. Lux values of 500 and 800 arehighlighted in red. These span the valuestypically recommended for reading andother activities (see below). As indicatedby the high Pearson R-squared valuesgiven inFigure 3, there is an excellentlinearity between the power setting andthe luminance value for each wall color.For these data, an R-squared value above0.980 (R value above 0.990) is signiﬁcantat the 0.01 level. This condition is met byall four experimental lines inFigure 3,conﬁrming that the relationship betweenlight intensity and power is linear overthis range. Linearity is expected as itindicates that the illumination eﬃcacy of alight does not change with power over therange we studied.9
►RESULTS AND DISCUSSION
We intentionally used an interior room,rather than one with a window, in thisstudy. A window would add complexityand confound the results. We wouldexpect less electricity use during theday, when the room beneﬁts from abright outdoors, but signiﬁcantly moreelectricity use at night, when most of thelight falling on the windows is lost to theoutdoors. The balance between savingsduring the day and losses at night is diﬃ-cult to quantify since it depends on theroom’s use (work or home), the lengthof the day, and the number of activehours during daytime and nighttime.In addition, many oﬃces do not havewindows—in a 2015 survey of oﬃceemployees, 61.2% reported that they didnot sit near a window and so had verylittle natural light.10
Our interest is in two lux levels: 500lux, which is speciﬁed by Europeanlighting standard EN12464-1:2011, and800 lux, which was preferred by 60%of European oﬃce employees in a 2015survey6and is recommended for workers45 years or older.11We found that theresponse of luminance to power level tobe linear in all cases (seeFigure 3for anexample), and so the percent extra elec-tricity needed to illuminate the room to500 lux, for a given room color and lamptype, is the same as the percent extraelectricity needed to illuminate theroom to 800 lux (that is, luminance islinear with electrical power).The luminosity of the lights at each ofthe four light settings was determined byplacing a photo meter directly below eachlight ﬁxture. This was measured for eachroom color, to conﬁrm that the radiationbeing measured was entirely from thelights themselves, with only an insignif-icant amount, if any, coming from thewalls. This was conﬁrmed to be the case.
Our analysis included all data points,but for brevity not all of these points willbe detailed in this article. Instead, wewill focus on a subset of results that webelieve to be representative of the entiredata set. This subset consisted of:
•Room center; meter oriented facingup; 4100K bulbs
•Room center; meter oriented at 45°;4100K bulbs
•Against the wall (“Wall 2” inFigure1); meter oriented facing up; 4100Kbulbs
•Against the wall (“Wall 2” inFigure1); meter oriented at 45° into theroom; 4100K bulbs
Our reason for analyzing these par-ticular locationsand orienta-tions is that webelieve them tobe relevant todiﬀerent usagesof the room. If theroom is used as aconference room,then the mostlikely location fora table would be the center of the room. If it is used as anoﬃce, then the most likely location for adesk would be along a wall. The two dif-ferent orientations (facing up and at 45°)are those used for observing an objectﬂat on the table or desk, or for holding abook at these locations.
Our chief interest in this work isto determine the relative amount ofextra energy needed to illuminate theroom to the same level for diﬀerentwall darkness. We consider the whiteroom results to be the baseline, withan assigned value of 1.0, and calculatehow much more energy, relative to thebaseline value, is required to illuminatethe room to the same level of brightnessas the white room. Averaged values arereported inTable 1and plotted inFigure4.
Note that one data point was omittedfrom our analysis—the room center, at45° orientation, for the light gray room—because it was anomalous.
Results for the room center, with themeter held horizontally, were the leastsensitive to wall brightness. This is expected since this is beneath the lights,and so most illumination will be directlyfrom the source. The wall location, fac-ing 45° into the room, was the next leastsensitive. Here the meter was pointednearly at the lights, but since thislocation is further away than the centerlocation, the electrical need is greaterthan for the center location. The locationand orientation most sensitive to wallbrightness was the wall, facing up. Thisis understandable since a signiﬁcantportion of light striking this surface willcome from the wall.*
Overall, we ﬁnd that, for the blackpaint, between 41% and 85% additionallighting is required compared to thewhite room. While this is an extremecase (rooms, especially those with nowindows, are seldom painted black),even for a light gray paint, between 21%and 31% additional lighting is requiredcompared to the white room. This isa signiﬁcant increase in electricityrequirements which, as detailed above,constitutes a signiﬁcant portion of theoverall electricity demand globally.
We will discuss the implications ofthese results to real world room colorsbelow.
Modeling was done using DIALux soft-ware.12This software is widely avail-able and heavily used by architects andbuilding designers.
While we would have liked to modeleach of the four room locations andorientations, this software is limited tomodeling only horizontal orientations.In addition, the minimum wall bright-ness was 0.10, rather than the 0.05 thatwe achieved experimentally. We, there-fore, modelled the horizontal center andwall locations at wall reﬂectance valuesof 0.90, 0.60, 0.20, and 0.10. The lightsources available for modeling did notmatch either of our two lamps, and so a32 Watt ﬂuorescent, 5208K source wasused in the model. This should have noimpact on our conclusions, since we areconcerned with analyzing energy needson a relative basis (relative to brightwhite) rather than an absolute basis.
As was done with the experimentalresults, we report, inTable 2, the relativeelectricity requirements at equal bright-ness, for the four wall colors, at each ofthe two locations. This data is showngraphically inFigure 5.
Comparing the results inTables 1and2graphically (Figure 5), we see that themodel agrees well with the experimen-tal results for the center reading, butis 8% higher than the experiment forthe measurement near the wall. This isseen in the slopes of the best ﬁt lines inFigure 5(seeTable 3). The slope of theline is a measure of the sensitivity of thebrightness at that location and orienta-tion to wall darkness. Assuming that ourresults are representative of most oﬃces,this suggests that architects may bespecifying brighter light intensities thanis necessary in some cases. We cannotoﬀer an explanation for this discrepancy,since we do not know the details of howthe model bases its calculations.
►IMPLICATIONS FOR PAINT CONSUMERS
Color is a leading consideration forinterior architectural (décor) paints, andcolor choice is quite often the ﬁrst ques-tion asked of a customer entering a paintstore. While statistics are not availableas to which colors are popular amongbusiness and home owners, the tradi-tional choice is an oﬀ white. However,many interior designers recommenddarker, bolder colors to their clients asa way to stand out and make a fashionstatement. This is reﬂected in the choiceof “color of the year” made by many inte-rior architectural producers. We havecollected 19 of these colors from the lasttwo years. Tristimulus Y values andcorresponding L* values for these paintsare given inTable 4.
As can be seen fromTable 4, the “colorof the year” paints are relatively dark(the average tristimulus Y value is 23.3and an average L* value of 49.8). Thisis consistent with many of their names,such as “Dark Navy” and “Deep Onyx.”
To estimate the lighting costs associ-ated with these colors, we have indicatedtheir reﬂectance values in Figure6 alongwith the best ﬁt lines for the four location/orientation pairs that we analyzed (fromFigure 5). We tabulate the additional elec-trical load for the four location/orientationpairs for our room inTable 4. We see fromthis table that the additional electricityrequirements for these paints vary from12% to 84% versus white walls. It is clearthat the additional electricity burden ofusing relatively dark paints on interiorwalls is very real.
►ECONOMIC AND ENVIRONMENTAL CONSEQUENCES
From our data we can calculate themonetary cost of lighting rooms withbright walls over rooms with dark walls.While electricity cost varies globally, wewill calculate these costs for one speciﬁcregion as an indicator of the value ofbright walls in general. For this calcu-lation we will use data from the UnitedStates, where the average user cost of akilowatt hour of electricity is $0.133.13We assume that the room is used ﬁvedays a week for nine hours per day.
The results of these calculations forour room are shown inTable 5, aver-aged over the two light types. Here weconsider two lighting levels (500 luxand 800 lux) and two measurementlocations (wall and room center, bothoriented up). The ﬁnancial cost of a darkroom, compared to a white room, issubstantial—over one year, the addi-tional electricity cost of the black roomvaries from $16.40 to $84.23, dependingon lighting level and where in the roomlight intensity is measured. Over theservice life of the paint, the additionalcost of lighting a black room far exceedsthe initial cost of the paint, and thecost of using greater amounts of TiO2tobrighten a room is quickly oﬀset by theelectricity savings.
In addition to the monetary cost of adark wall, there is an important envi-ronmental cost. This cost is the amountof CO2released to the environment,both in making the TiO2pigment and ingenerating the electricity consumed bythe lights. We can quantify this cost interms of a payback period—how long ittakes for the CO2savings from using lesselectricity to oﬀset the CO2generatedwhen making the TiO2white pig-ment. For this calculation, we assumethat each kilowatt hour of generated electricity results in the release of 0.71lb of CO2,14that the white paint contains2.5 lb of TiO2per gallon, and that pro-duction of each pound of TiO2results inthe release of 5.0 lb of CO2.
As seen inTable 5, this environmentalpayback period is quite short—a littleover eight weeks for the least sensitivelocation (room center) at lowest illumi-nation target, and less than two weeksfor the most sensitive location (at thewall) and highest illumination level.
►COMPARISON TO WHITE ROOF ENERGY SAVINGS
Over the years there have been numer-ous initiatives to decrease electricalconsumption, both to reduce cost andto protect the environment. One suchprogram is the cool roof (or white build-ing) initiative. The concept is straight-forward—by replacing dark roofs withlight roofs, much solar energy can bereﬂected away from a structure, ratherthan be absorbed in to it in the form ofheat. This could signiﬁcantly decreasethe cooling burden on the structureduring hot summer months.
Note that this not only results in lessoverall electricity use, but it does soduring peak electricity demand times,since in many cities the greatest elec-trical use occurs on the hottest days ofthe year. By decreasing peak demands,the number of power plants neededto service a city will be fewer, sincethis number is determined by the peak(maximum) electricity requirement,rather than the average requirement.
An estimate of the electricity savingsgained by increasing the albedo of roofsfrom a solar reﬂectance value of 0.2 toa value of 0.6, in 27 cities around theglobe, was reported in 2007.15Thesesavings are, of course, greater in cities characterized by very hot summers thanin more moderate climates. Overall,the summer electrical savings wereestimated to be between 11% and 75%for this level of brightness increase.These estimates are quite comparableto the savings potential we estimatefor painting rooms with bright colors,reinforcing the consequences of using adark interior color to overall electricityconsumption.
Our work quantiﬁes, for our test room, therelative energy required to meet lightingtargets as a function of wall reﬂectance.While we studied only one room, webelieve the conclusions we draw from it tobe valid for similar rooms, at least at thesemi-quantitative level.
We found that the amount of electric-ity required to light the room to 500 lux,the minimum level generally acceptedas adequate for oﬃce work, and 800 lux,the level preferred by many, correlatesvery strongly with the brightness ofthe room paint. This energy demandwas linear over the range we studied.Electricity requirements increased byas much as a factor of 1.85 between thebright white and black paints.
While this is an extreme brightnessrange (very few rooms are expected tobe painted black), some “colors of theyear,” as chosen by paint manufacturers,are quite dark. For the colors that weexamined, we estimate the additionalelectricity requirements for them, rela-tive to the requirement for a white wall,to vary by factors of 1.12 to 1.84. Thesevalues are quite signiﬁcant—as a pointof reference, depending on location, coolroofs are estimated to reduce coolingcosts by between 11% and 75% whenthe reﬂectivity of the roof is increasedfrom 0.20 to 0.60. On a percentage basis,these savings are quite comparable tothe electricity savings we demonstratedfor a bright white room compared to aroom painted with a “color of the year”(between 12% and 84%).
The increase in electricity consump-tion for dark rooms comes at a cost—both ﬁnancial and environmental. Theannual additional cost of lighting a darkroom ranges, for our room, from $16.40to $84.23, depending on how the room is used (oﬃce or conference room). Asfor environmental considerations, ouranalysis showed that there is a very fastbreakeven time (a few weeks) for boththe monetary cost of the TiO2used inthe white paint and the CO2released inthe production of the TiO2.
We also found that a popular archi-tecture modeling program overes-timated the eﬀect of wall color onlighting needs by up to 8%. Relying onthis model alone would lead to brighterlighting than necessary, but at theexpense of additional cost and environ-mental burden.
While personal preference willalways be a major factor in color choice,it is important that the consumerbe aware of all costs of a given coloroption—not only in terms of monetarycosts (in the form of higher electricitybills), but also environmental costs(emissions from power plants). Thoseconsumers wanting a “green” paint mayﬁnd the color they are looking for is, infact, white.