Open access peer-reviewed chapter

Air Change Measurements Using Tracer Gases: Methods and Results. Significance of air change for indoor air quality

By Detlef Laussmann and Dieter Helm

Submitted: October 28th 2010Reviewed: March 22nd 2011Published: July 27th 2011

DOI: 10.5772/18600

Downloaded: 3021

1.Introduction

Both comfortable and healthy indoor climate conditions can only be achieved by constant fresh air supply. However, the minimum of air change required to reach this goal is depending upon different perspectives. Measures for thermal insulation and energy savings are difficult to bring in line with air quality requirements that result from findings of epidemiological studies (Seppänen et al., 1999; Seppänen& Fisk, 2004; Wargocki et al., 2002), and supply air facilities considering the technical construction of buildings and safety aspects (Erhorn&Gertis, 1986). Moreover, construction deficits impairing the integrity of the building envelope, meteorological conditions (thermal and flow induced pressure differences) and, not least, the behaviour of the residents, affectthe air change in a variety of ways (Heidt, 1987).

For a number of different reasons it is necessary and desirable to examine the real fresh air flow between indoor and outdoor climate under given circumstances. Air change processes are of particular importance in studies focusing on their relationship to indoor air pollutants.

For manifold reasons the air quality inside buildings has been intensively investigated since quite a long time. Of particular relevance are the following aspects (Seifert &Salthammer, 2003):

  • In countries with a cold or temperate climate, inhabitants spend more than 50% of the time in their homes.In certain population groups (e.g. infants and the elderly), this proportion is even exceeding 90%.

  • Indoor air contains a wide range of different organic and inorganic components. Therefore, the carbon dioxide concentration alone cannot always be regarded as an indicator of air quality (Fanger, 1988; Persily, 1997). Above all, organic compounds which are released from the building, furnishings, household and hobby devices, as well as by daily activities of the inhabitants, such ascooking, baking or frying, and especially smoking, altogether contribute to air pollution caused by volatile organic compounds.

The energy crisis in the 1970s caused an increase in energy costs, evoking an urgent need to reduce the consumption of heat energy. The preventionof heat loss in homes is a very effective way to save energy and related costs. For climate protection and the reduction of global CO2 emissions the economical use of energy resources is of outstanding significance. Thermal insulation of the building envelope protects efficiently against heat loss, but enhanced tightness includes the disadvantage of reduced air change, thereby increasing the indoor pollutant concentration. To evaluate the indoor air quality under conditions of natural ventilation, the air change rate (ACR) can be determined through standard tracer gas measurement.

For this purpose, a small amount of the tracer gas is released in the room (or building) under study and its concentration is then recorded as a function of time. Subsequently, using appropriate evaluation algorithms the ACR can be calculated from the data obtained.

The first sectionis introducing basic physical principles of air change processes between indoor and outdoor environment, followed by a description of tracer gases which are frequently used in daily routine, and methods determining the ACR. Moreover, the applicability of carbon dioxide as a tracer gas has been compared with results obtained by hexafluorobenzene or sulphur-hexafluoride. In addition, we are discussing the impact of weather conditions on ACR data obtainedunder natural ventilation. This section is followed by a review on ACR due to passive ventilation through facades in selected residences in Berlin. Finally, the relationship between air change and the concentration of selected volatile compounds under worst case conditions will be discussed.

2. Modelling air change in indoor rooms

According to VDI (2001) 4300, part 7, air change is defined as the ratio of air supply Q(t) into a zone (i.e. a room or space) in relation to the volume of this zone VR (room volume) and is generally expressed as air change per hour [h-1] or [ACH]. The following equation expresses this definition:

λ(t)=Q(t)/VRE1

λ(t) is the ventilation rate or air change rate [h-1],

Q(t) is the air supply into a room [m³/h],

VR is the room's volume [m³], and t = time [h].

2.1. Model assumptions

For the model described here, which is designed to calculate the time course of the tracer gas concentration, the following simplifying assumptions has been made:

The tracer gas is considered to be chemically stable and inert; i.e. there will be no chemical reactions capable to alter the concentration of the tracer gas in the room.

There will be no adsorption processes on walls, ceiling or furnishings of the room that may lower the concentration of the tracer gas in the room.

The air is considered to be completely mixed throughout the measurements. Inside the room there are no concentration gradients, i.e. the concentration of the tracer gas at a given time is the same for the whole room.

An exchange of tracer gas-containing air with ambient air only occur in those areas that are in contact with the outside, i.e. air change with other interior spaces is considered to be negligible. The room in which the tracer gas was released is considered a single zone system.

The exchange processes that take place during the measurement period are assumed to be temporally invariant. The air supply rate Q(t) and, thus, the air change rate λ(t) are constant. Q(t) and λ(t) can be replaced by Q and λ.

2.2. Model equations

The basis for the description of the relationship between the mass or concentration of a gaseous substance in a space as a function of time is the mass balance equation. This equation expresses that the mass and – in a fixed volume – the concentration of a tracer gas can only change when either more tracer gas is added to the original amount or tracer gas is removed by elimination processes. Considering the above assumptions, the following supply and removal processes are significant for the tracer gas concentration in the room air (physical dimensions of these variables are given in brackets):

  • Transport of tracer gas from the room air to the outside: Q*Ci [mass per time unit]

  • Transport of tracer gas from the outside air into the room air Q*Ca [mass per time unit]

  • (Constant) emission E of tracer gas into the space by a tracer gas source [mass per time unit]

Thus, the mass balance equation can be formulated as the following differential equation (Heidt& Werner, 1986): (equation 2a)

After dividing both sides of the equation by the volume VR, we obtain an ordinary differential equation which describes the concentration change of a tracer gas in the room per time unit: (equation 2b)

As stated above, Q/VR is defined as the air change rate. When Q/VR in Eq. 2b is replaced by λ (Eq. 1) we obtain Eq. 2c:

a)VR*dCi(t)/dt=Ci*Q+Ca*Q+E          or               VR*dCi(t)/dt=(CiCa)*Q+Eb)dCi(t)/dt=(CiCa)*Q/VR+E/VRc)dCi(t)/dt=(CiCa)*λ+E/VRCa: tracer gas concentration in outside air [mass / volume]

Ci: tracer gas concentration in the indoor air [mass / volume]

Q: exchange air flow between room and outside [volume / time unit]

E: amount of tracer gas emitted per unit time [mass / time unit]

VR: room volume

t: time

Equation 2c expresses that under constant homogeneous mixing the concentration change of the tracer gas is proportional to the concentration difference between indoor and outdoor spaces (Ci – Ca) at time t, the air change rate λ, and the amount of tracer gas emitted per time unit (emission rate E).

By integration we obtain the starting conditions at time 0 (C(t=0) = C0):

a)Ci(t)=Ca+E/(λ*VR)+[C0CaE/(λ*VR)]exp(λti)Expandingandtransposingyields:b)  Ci(t)=(C0Ca)exp(λti)+Ca+E/(λ*VR)[1exp(λti)]c)  Ci(t)=C0exp(λti)E2

Integration of the differential equation thus leads to an exponential function with the air change rate λ in the exponent. Equation 3b is the basis for the mathematical analysis of tracer gas measurements that are recorded as concentration-time curves. This function is characterised by the following features: If no tracer gas is emitted into the room (i.e. E = 0), and there is already a non-zero tracer gas concentration C0 present at the time t = 0 which is higher than the outdoor air concentration Ca, then the expression (C0 – Ca) exp(-λt) + Ca describes the elimination of the tracer gas out of the room. The curve starts with the initial concentration C0 and decays exponentially until the ambient tracer gas concentration Ca or any other constant background concentration is reached. If a tracer gas is used, which does not occur in the outside air (i.e. Ca = 0) the concentration decreases over time to the value zero. Equation 3b can be simplified to Equation 3 c:

If a source in the room is emitting tracer gas with a constant rate E and if the initial tracer gas concentration in the room at the beginning of the measurement C0 is equal to the outside concentration Ca (C0 = Ca), then the tracer gas will accumulate within the room. Starting with the concentration Ca, the mathematical function grows with increasing t asymptotically towards Ceq, which reaches the final value of Ca + E/(λ*VR) at infinity.

a)  C(t)=Ca+[E/(λ*VR)][1exp(λt)]Whentheoutdoorairconcentrationcanbeneglected,then:b)  C(t)=[E/(λ*VR)][1exp(λt)]Fort>>1/λEquation4bcanbesimplifiedto:c)  Ceq=E/(λ*VR)E3

Ceq: equilibrium concentration

Equation 4c reflects the fact that under equilibrium conditions (i.e. emission equals elimination) the tracer gas concentration Ceq is proportionally dependent on the emission rate E but inversely proportionally dependent on both the air change rate λ and the volume of the room VR. This relationship forms the basis for the determination of the air change rate with tracer gas measurements made under equilibrium conditions.

3. Determination of the air change rate (ACR) using tracer gases

In addition to the physical properties discussed in Section 2.1 (model assumptions) tracer gases should fulfil some other requirements in regard to their practical suitability. These include health safety aspects, low environmental burden, high availability and good handling in practical use at the lowest costs possible and, not least, tracer gases should be well recordable with established measurement techniques over a wide concentration range and with high selectivity (Raatschen, 1995).

3.1. Requirements for ideal and commonly used tracer gases

In the past, a number of gaseous substances such as helium, hydrogen, oxygen, carbon monoxide, methane, acetone, and the radioactive noble gases argon-41 and krypton-85, have been studied and tested for the determination of ACR, mostly in comparison with other tracer gases. Reviews are given by Grimsrud et al. (1980), Shaw (1984), and Sherman (1990). Until the early 1990s, krypton-85 was still used for air change measurement (Schulze &Schuschke, 1990), but later skipped for safety reasons (radiation protection). Nowadays, the following tracer gases are mainly used in practice (Raatschen, 1995):

  • Nitrous oxide (N2O)

  • Sulphur hexafluoride (SF6)

  • Halogenated hydrocarbons, such as hexafluorobenzene (C6F6) and perfluorocarbons (PFC).

3.1.1. Nitrous oxide

Formerly, nitrous oxide (N2O) was widely used for air change measurement in buildings, primarily in Europe (Heidt& Werner, 1986; Keller &Beckert, 1994; Salthammer, 1994; Wegner, 1983, 1984). In the U.S. it was rarely used because of its low TLV (threshold limit value) of 50 ppm (Lagus& Grot, 1997). Germany's equivalent of TLV, the MAK, is however 100 ppm. For precautionary reasons it should not be used in occupied buildings (Raatschen, 1995). Other disadvantages of N2O are the ease of adsorption on surfaces at concentrations below 1000 ppm, and its high solubility in water, which means that the air change rate can be substantially overestimated in very airtight rooms (Schulze &Schuschke, 1990).

3.1.2. Sulphur hexafluoride (SF6)

After Gregory's observation in 1962, that sulphur hexafluoride (SF6) can be measured reliably on the nanogram scale with electron capture detection – ECD (Gregory, 1962), it has been widely used for air infiltration measurement in buildings since the early 1970s (Drivas et al., 1972; Hunt & Burch, 1975). Of all candidates, the characteristics of SF6 are nearest to the ideal of a tracer gas. Today, SF6 is the most frequently used tracer gas worldwide, which is confirmed by the number of relevant publications. In this chapter, only on a very limited selection of publications could be considered (Chuah et al., 1997; Howard-Reed et al., 2002; Kumar et al., 1979; Lagus& Grot, 1997; Raatschen, 1995; Shaw, 1984; Walker & Forest, 1995; Wilson et al., 1996). SF6 is very stable and only decomposes above 550 C. Background concentration of SF6 in ambient air is ≈1 ppt(6 ng/m3) (Raatschen, 1995). Although its density is about five times higher than that of air, this difference causes no systematically distorting effects on the results of air change measurements with concentrations usually applied in practice (Niemelä et al., 1991; Shaw, 1984). SF6 can be used in occupied buildings. It can be recorded with high accuracy within a wide concentration range. According to Raatschen (1995) the concentrations commonly used for indoor air change measurement are not exceeding one hundredth of the German MAK value which was defined to be 1000 ppm or 6100 mg/m3 (TRGS 900, 1999). It should be noted that this value is not a toxicity limit. It is simply defined as the still manageable analytical upper limit for gases which are not imminently toxic (BIA-Report, 2001). Due to its high stability SF6 is only very slowly degraded in the atmosphere and it belongs, like the perfluorinated hydrocarbons, to the climatic relevant greenhouse gases. Therefore, to avoid unnecessary environment hazard it should be used carefully and sparingly in concentrations as low as possible, like other tracer gases.

3.1.3. Hexafluorobenzene (C6F6) and perfluorocarbon-hydrocarbon tracers (PFT)

Hexafluorobenzene, and perfluorocarbon-hydrocarbon tracers (PFT) such as perfluorodimethylcyclobutane, perfluoromethylcyclobutane, perfluorodimethylcyclo-hexane, and perfluoromethylcyclohexane are also appropriate as tracer gases and are preferably used in the determination of ACR with the constant injection method applying the passive sampler technique. Enrichment for active sampling is also possible (Cheong &Riffat, 1995; Dietz & Cote, 1982; Dietz et al., 1986; Krooß et al., 1997; Mailahn et al., 1989; Salmon et al., 2000). Since these compounds, unlike SF6, adsorb well on activated carbon or Tenax they are frequently used in field studies (surveys) to determine indoor air change rates (Andersen et al., 1997; Bornehag et al., 2005; Hirsch et al., 2000; Lembrechts et al., 2001; Øie et al., 1997, 1998; Parker, 1986; Pandian et al., 1993; Ruotsalainen et al., 1992; Sakaguchi&Akabayashi, 2003). The disadvantage of these compounds, however, is that they tend to attach to room surfaces and the emission rates of these gases are strongly temperature dependent (Hill et al., 2000). Thus, a sufficiently long conditioning period is necessary as well as an accurate temperature control of the storage vessels.

3.2. Methods for the determination of ACR

Basically, three appropriate methods exist for the determination of ACR using tracer gases. According to VDI (2001) 4300, part 7, these are the concentration decay method, the constant injection method, and the constant concentration method.

3.2.1. Concentration decay method

Tracer gas is injected into the room for a short period of time, either from a gas bottle with pressure reducer or manually from filled gas tanks. After mixing with the room air the tracer gas concentration is measured at regular time intervals. Because the decay curve of the tracer gas concentration C follows an exponential course when completely mixed with the room air (see Eq. 3c) each sampling will not only remove old air but also a certain amount of fresh air supply as well. This essentially means that there will be still ≈37% (100/e1) of the originally added tracer gas (37% old room air), after a complete air change cycle has occurred. The time (1/λ), after which the air change cycle is completed is known as the nominal time constant τ (Maas, 1997; Sherman, 1990). After 3τ (3/λ), 4τ (4/λ) and 4.6τ (4.6/λ), the tracer gas concentration in the room volume under study is 5%, 2% and 1% of its initial value, respectively (Fig. 1a).

By using the concentration decay function

C(t)=C0*eλ*ta)  lnCi(t)=lnC0λ*(ti)b)  [lnCi(t=ti)lnCi(t=ti+1)]/(ti+1ti)=λ.E4

Ci(ti) is the tracer gas concentration at time ti

ti+1 – ti is the time interval between two measurements

and applying non linear regression analysis we can determine the ACRλ (C0: tracer gas concentration at time t=0 (Sherman, 1990)). However, in most cases logarithmic concentration values are used to obtain a linear relationship between the logarithm of the tracer gas concentration Ci(t) and the time t (Fig. 1b): (equation 5a)

The ACR λ is then calculated via linear regression analysis according to Eq. 5a. Both evaluation options – the linear and the non-linear regression analysis – are particularly well suited for the examination of concentration-time curves recorded over a longer period of time. If only a few measurements are available, e.g. when sampling is performed with syringes or other appropriate devices, then the ACR can be determined using the following relationship: (equation 5b)

Figure 1.

Relationship between tracer gas concentration and time; a, untransformed ordinate; b, after logarithmic transformation.

An illustration of the procedure is given in Fig. 2a. Sulphur hexafluoride (SF6) was released in an office space (volume 48 m³, equipped with two double glazed box windows). Initial concentration of the tracer gas was adjusted to ≈5 mg/m³. The concentration was measured for 7 hours in 15 minutes intervals using a photo-acoustic infrared detector with selective filter (Bruel&Kjaer Single Gas Monitor). During the measurement the office was unoccupied; door and windows were kept closed. Tracer gas and room air were not continuously mixed with a fan or similar device. For the analysis, SF6 concentration values were chosen in such a way that each concentration-time pair – c(ti+1), c(ti) – covered exactly a time interval of one hour, meaning that the denominator was equal to 1 when determining the air change rate (ACR). After choosing the six measurement points depicted in Fig. 2a we calculated an average ACR of 0.327 h-1. The standard deviation was 0.038 h-1, which corresponds to a coefficient of variation of ≈12%. When the same measuring points were analysed by linear regression analysis we obtained an ACR of 0.323 h-1 (Fig. 2b). Both values are almost equal. The determination of ACR using concentration-time pairs is also known as the two-point method (Sherman, 1990). For this method, a number of at least five concentration-time pairs are recommended which should be evenly distributed over the entire measurement period (ASTM E 741 – 00; VDI, 2001 ). In order to reduce the statistical error to ≤10% the time interval between the first and the last measuring point should be in the order of magnitude of the nominal time constant (ASTM E 741 – 00; Heidt& Werner, 1986; Maas, 1997; Sherman, 1990). This prerequisite is, however, hardly fulfilled for ACR between 0.1 and 0.2 h-1 or even lower, since the required minimum decay time would be between 5 and 10 hours or more. Most probably, in the case of very low ACR a statistical error of more than 10% must be accepted. Recommended minimum time intervals between the first and the last measurement and the corresponding measurement intervals are summarised in Table 1, according to ASTM E 741 – 00 and VDI (2001).

Figure 2.

ACR determination from concentration decay with the two-point method, time interval ti+1 – ti between two measurements: one hour; a, linear scale; b, logarithmic scale, linear regression.

Table 1.

Examples for minimum time spans between first and last sampling based on recommendations of ASTM E 471 – 00 and recommended sampling intervals (VDI, 2001) for air change measurement with the decay method.

The concentration decay method is the most commonly used one in practice. It is particularly well suited for the determination of ACR up to 10 h -1 (10 ACH) in indoor rooms with a volume below 500 m3 (VDI, 2001). If data loggers are used to record the concentration decay, the measurement intervals can be shortened down to seconds and in consequence ACR >10 h-1 can be determined.

3.2.2. Constant injection method

To determine ACR with this method a diffusion tube containing C6F6 or PFT is frequently used as tracer gas source. A defined amount of tracer gas is constantly emitted over a certain period of time. Thus, the tracer gas concentration increases with time and reaches a stable value (equilibrium concentration) which depends on the room volume VR, the air change rate λ, and the emission rate E. At that time point, one or more air samples are taken and the tracer gas concentration is determined for each sample. The ACR can be calculated after solving Eq. 4c for λ. Fig. 3a depicts an example of the constant injection method.

Crucial for this method is, however, that sampling can only be started when the tracer gas concentration is near the equilibrium. If sampling starts too soon, an overestimation of the ACR will be the consequence. The time to reach approximate equilibrium conditions depends on λ (see Fig. 3b).

Figure 3.

Determination of ACR; a, constant injection method; b, concentrations after constant injectionfor different ACR.

According to ASTM E 741 – 00, the tracer gas concentration should have reached at least 95% of the equilibrium concentration before measurements can be started. The time until approximate equilibrium is reached can be estimated applying Eqs. 4b, 4c. However, under worst case conditions in rooms with very low air changes, it will be difficult to correctly measure ACRs with this method, because the build-up time to reach nearly equilibrium concentrations is too long (Table 2).

The time intervals for the tracer gas concentration to reach 95%, 98% or 99% of the target equilibrium value (Ceq) are 3/λ, 4/λ and 4.6/λ, respectively (cf. Table 2).

As Table 2 shows, the time periods needed to reach approximate equilibrium can be several days when the ACR is only 0.1 h-1 or even lower. For this reason, the constant injection method is suitable only for short-term measurements of ACR above 0.3 h-1 when diffusion tubes are used as tracer gas source. Under usual "worst case" conditions (i.e. last active window ventilation 10-12 hrs before sampling) and ACR <0.3 h-1 the tracer gas concentrations are still too far away from the equilibrium. The advantage of the constant injection method is that it can be used for indoor hygiene studies addressing the relationship between pollution and air change in daily used rooms. In this case, tracer gas measurements can be performed with passive samplers and exposure times range from days to weeks.

Table 2.

Time spans [hours] needed to reach 95%, 98% or 99% of equilibrium concentration (Ceq) for a wide range of air change rates.

3.2.3. Constant concentration method

During constant and thorough mixing with the indoor air, tracer gas is released in the room until a predefined concentration is reached. During the entire measurement the tracer gas concentration is kept constant with an automated dosing and control system. Under the condition of constant tracer gas concentration the air supply is proportional to the tracer gas supply rate. The air supply rate can then be calculated from the ratio of the tracer gas supply to the tracer gas concentration. If the room volume is known, the ACR can be calculated from this ratio (Chao et al., 2004; Kumar et al., 1979; Maas, 1997). An advantage of the constant concentration method is that even short-term changes of air supply can be detected. Compared to the previously described methods the technical equipment required for this method is, however, rather expensive and thus this method is comparatively rarely used for indoor air quality evaluation.

3.3. The use of carbon dioxide as a tracer gas

Carbon dioxide (CO2) is one of the gaseous organic compounds always detectable in the indoor air. Since humans exhale metabolic carbon dioxide in considerable quantities, its concentration can increase to several thousand ppm (ml/m3 room air) within a short time. CO2 concentration is often used to assess the air quality of occupied rooms. In this context we remind of Pettenkofer's reference concentration. Already in 1858 the German chemist and hygienist pointed out that a CO2 concentration of 1000 ppm (0.1 vol %) is the upper tolerable limit in indoor environments. Nowadays CO2 measurements are often used for the determination of the indoor ACR, because it can be easily quantified and the required devices are reasonably priced and easy to operate. Moreover, CO2 fulfils a number of the above mentioned specifications of a good tracer gas.

A huge number of studies are published testing the feasibility of exhaled human carbon dioxide as tracer gas in air change settings. Next to the already mentioned work of Pettenkofer (1858) we would like to allude to the studies done by Penman (1980), Penman & Rashid (1982), and Smith (1988) which are of special importance. Results of Dols&Persily (1992), Nabinger et al. (1994), and Persily (1997) have, however, demonstrated that ACR cannot be reliably determined from spot, peak or average values of the CO2 concentration inside buldings, because these values are strongly influenced by the number of occupants in the rooms, their times of stay, and, hence, the incessantly changing carbon dioxide supply rates. Depending on the amount of natural ventilation the ACR is sometimes over-estimated up to 2-fold of the real value. The reason is that the air-tightness of modern buildings and the usual sojourn times of the occupants prevent in most cases that the equilibrium concentration can be approached. In practice it is much better to derive the ACR from the decay or build-up curve as was already shown (Barankova, 2005; Bekö et al., 2010; Chao et al., 1997; Chung & Hsu, 2001; Guo& Lewis, 2007; Menzies et al., 1995; Roulet&Foradini, 2002; Schulze &Schuschke, 1990; Sekhar, 2004; Shaw, 1984).

3.3.1. Statistical evaluation of CO2 decay curves (with examples)

After CO2 is released in a room, either as exhaled breath or via a gas container, its concentration will decay exponentially, if no further CO2 supply occurs. To exemplify this, the CO2 decay recorded in a bedroom of an older building (built 1908) is depicted in Fig. 4. In 1990 this bedroom was equipped with a double box window; the room volume is 30 m³. The room was doped by a person with exhaled carbon dioxide, then the measurement was started and the room was left. During the entire procedure the room was unoccupied – door and window were kept closed. Measurement of CO2 concentration was done by a CO2 sensitive probe with infrared absorption. A Testo 400 device (Testo, Lenzkirch, Germany) was used for data logging and as control unit.

Unlike other tracer gases, CO2 has the particularity that there is always a certain amount of CO2 in the outdoor air meaning that the background concentration cannot be neglected. In general the outdoor air concentration of CO2 is between 350 and 450 ppm or even higher, depending on the season. Thus, the CO2 decay curve will not decline to zero. Instead, the CO2 concentration decreases to values which are near to that of the outdoor air. This must be taken into account when analysing CO2 decay curves. Eq. 5 cannot be applied; the air change rate (ACR) must be determined using the term derived from Eq. 3 b:

C(t)=(C0Ca)*eλ*t+Caa)  ln(C(t)Ca)=ln(C0Ca)λ*tE5

C0: initial concentration, Ca: concentration in the ambient air (i.e. background concentration)

It is possible to examine the linearised curve when the background concentration is subtracted before linearising. Therefore, the background concentration must be determined by an additional measurement. (equation 6a)

With this relation the ACR can be determined by linear regression analysis. In practice, however, this will be difficult in most cases since it is not feasible to record the CO2-concentration until the decay curve reaches background level. Modern buildings are in general designed to achieve high airtightness resulting in a very slow decay, and hence the linearisation of the decay curve using the logarithm of the concentration differences is not possible. Furthermore, the background concentrations are often unknown or not accessible to direct measurement. In these cases, the air change rate λ can be determined only by non-linear regression with iterative calculation methods according to the model as given in Eq. 6. Iterative calculation methods start with initial values (raw values) specified by the user for the function parameters, which are then improved iteratively by using the method of least squares until the model function is fitted optimally to the measured curve.

Figure 4.

Decay curves of CO2 (black glyphs) in a bedroom equipped with double box window and located in a building built in 1908, estimated with nonlinear regression (white curves).

Table 3.

Air change rates in a bedroom located in an older building (built in 1908) equipped in 1990 with wooden framed double box window, decay method and CO2 as tracer gas. Avg.: Average, CI: confidence interval

The results obtained from the CO2 decay curves depicted in Figure 4 were evaluated using the method of nonlinear regression and are shown in Table 3. The calculated ACR ranged from 0.25 to 0.4 h-1 (mean 0.32 h-1). The goodness of fit of the regression model is very high, since more than 99% of the variability of the CO2 concentration can be explained by the respective regression functions. This is also reflected by high accuracy of the estimates for the various ACR. The 95% confidence intervals are very narrow and deviate only about 2 – 3% downward and upward from each individual value. These results were due to both the low variance of the CO2 values from the fitted curves (white curves in Fig. 4) and the extensive measurement periods (several hours) which yielded high numbers of nodes (concentration-time data pairs) at measuring intervals of 3 minutes. However, the differences between day-to-day measurements are much larger. The 95% confidence interval for the mean ACR ranges from 0.226 to 0.411 h-1. A possible explanation for this relatively high day-to-day variation is given below (see section 3.5). ACR, which were determined under the terms of exclusive joint ventilation (i.e. windows and doors closed), vary considerably and show extremes that span two orders of magnitude. This is shown in Figures 5a and 5b where six examples of CO2 decay curves are depicted which can typically be recorded indoors. For illustrative reasons decay curves were selected which start at about almost the same initial concentration of 1500 – 1800 ppm and were recorded over a measurement period of more than 10 hours (Fig. 5a). Another selection criterion was the existence of stable weather conditions over the entire measurement period. Both figures demonstrate that CO2 decay curves can decrease exponentially over a period of 50 hours and longer under stable weather conditions (Fig. 5a).

Figure 5.

CO2 decay curves (black) in 6 rooms of different air tightness. White curves show results of regression analyses; a, nonlinear regression; b, linear regression of logarithmic values

This is an indication that tracer gas and room air can remain homogeneously mixed over such long periods without the need of a fan. The ACR estimated from the decay curves are given in Table 4, and the analysis of the curves was done with non-linear regression analysis (see Eq. 6). In addition to the air change rate λ, the background concentration Ca was also determined with the regression model and is presented in Table 4. After subtracting the respective background concentration from the measured CO2 concentration values, the logarithmic plot of the concentration differences shows a nearly perfect linear concentration-time relationship (Fig. 5b).

Table 4.

ACR and background concentrations with their 95% confidence intervals (95% CI) measured in 6 rooms with differing natural ventilation (windows and doors closed); estimated from CO2 decay curves with the statistical method of nonlinear regression (time interval between two concentration values: 3 min).

3.3.2. Determination of ACR from the concentration increase of carbon dioxide

When CO2 is supplied at a constant rate, it is possible to determine the ACR from the increasing concentration values (Figure 6). Evaluation of the concentration curve is the same as with the constant injection method. The air change rate λ is calculated with the non-linear regression model approach y = a0 + a1*[1-exp(-a2*t)] (cf. Eq. 4a), where a0 = Ca, a1 = E/(λ*VR), and a2 = λ. When the ACR is small, the build-up curve will approach the equilibrium only slowly, and this will lead to a quasi-linear concentration curve if the measurement period is too short. Therefore, the ACR calculated from this curve can be afflicted with large uncertainties. For this reason, to obtain a more reliable estimate of the ACR the concentration build-up must be recorded over a period as long as several hours, with intervals of a few minutes. When metabolic CO2 is used, measurements should be done best during sleep (of the occupants), because this comes nearest to the requirement of a constant CO2 supply. Figure 6 also shows the ACR determined from the analysis of the CO2 respective build-up curves depicted here.

These CO2 build-up curves were recorded in the same room as the decay curves in Figure 4. The measurement of the CO2 increase occurred, however, four years before, when the room was occupied by 3 persons (two adults, one child). The ACR of both series of measurements, decay and build-up curves, differ only slightly from each other taking daily variation into account.

Figure 6.

CO2 concentration build-up curves in a bedroom equipped with a double box window and located in a building built in 1908. Corresponding ACR, estimated with nonlinear regression (fitted values: white curves). Constant-injection method, room volume 30 m³, occupancy: 2 adults, one child.

3.4. Comparison of different tracer gases for the determination air change rates

The above examples have shown the applicability of determining ACR from CO2 concentration curves, when either the influence of the background concentration can be mathematically eliminated or a constant supply of CO2 is provided. It can, however, not be concluded that conventional and established tracer gases are dispensable. Compared to CO2 these have some indisputable advantages, namely a much wider range of possible applications and virtually negligible background concentrations.Both of which greatly simplifies the determination of ACR.

Under certain circumstances the use of established tracer gases is not possible, and in these cases we have to resort to the CO2 method. This may be necessary in patients with environmentally related health problems, who will not accept the use of tracer gases in their homes because they fear health hazards. In such cases, the CO2 method is a valuable alternative. It is therefore necessary to examine the extent to which results obtained with CO2 differ from those obtained with conventional and established tracer gases. To achieve this we performed a number of comparative studies by which an established tracer gas like hexafluorobenzene (C6F6) or sulphur hexafluoride (SF6) was used in parallel to CO2.

3.4.1. Determination of air change rates with CO2 and C6F6

The measurements described below were done in two different rooms, firstly, in a bedroom (24 m3 effective volume) of a terraced house (built December 1997) equipped with double glazed windows and insulating rubber seals, conducting measurements over a period of 5 days, and secondly, in an office space (72 m3 effective volume) of an old brick building (built ≈1900) equipped with two double-box windows. Windows and doors were closed during the experiment. A total of two measurement cycles were performed at intervals of two months.

At the start of the conditioning phase a diffusion tube which contained hexafluorobenzene was placed in both rooms. First sampling on Tenax-tubes occurred about 72 hours after the windows and doors were closed, further samples (two samples each) were taken every 24 hours. Room air was collected using sampling tubes (1 litre per tube) with a bellows pump, type Accuro (Dräger, Lübeck, Germany). ACR were calculated from the known emission rate of the C6F6-tubes, the equilibrium concentration and the effective room volume, according to the procedure for the constant injection method as described in VDI (2001), using Eq. 4c.

In the bedroom, CO2 concentration was measured continuously over the entire study period with measuring intervals of 3 minutes using the Testo 400-device equipped with CO2 probe, and the measured values were stored in the internal data logger. In theofficespace, the CO2 concentration was measuredwith asecond devicein a 24hour rhythmaccording to the American standard ASTMD6245-98. Measurements occurred mainly at nighttime in the then unoccupied office building. Prior to the start of the measurement, the CO2 concentration in the office space was set to ≈600 ppm by the experimenter. Thecalculation of the ACRwas achieved with non-linear regression analysisusingthestatisticalpackageSPSS, wherethemeasuredconcentration values were fitted to exponential terms with consideration of the calculatedbackgroundconcentrations.

Results of the CO2 measurements:

The CO2 time-concentration curve recorded for the bedroom is depicted in Figure 7. The bold black line represents the measured concentration values, whereas the grey line describes the graph which was obtained by means of section-wise non-linear regression analysis. Above the individual curve sections, the calculated ACR are plotted. Given the high air-tightness of the bedroom ACR as low as 0.05 to0.1h-1 were determined. It is obvious that the CO2 concentration increases very quickly during the night (one adult) to levels greater than 4000 ppm and decay exponentially during the day to values of 1500 - 2000 ppm. Due to thelowACR, the equilibriumconcentrationofCO2isnotcompassed during a 7 hours' sleep. The statistical analysis of the CO2 curves obtained for the office space equipped with double box windows resulted, however, in significantly higher ACR, which ranged between 0.2 and 1.4 h-1 because of different window and door features.

Results of the hexafluorobenzene measurements:

In Figure 8, the values of the hexafluorobenzene concentrations from a series of measurements in both rooms are shown together with calculated curves. The solid lines represent the concentration curves which are calculated using Eq. 4b on the basis of the known emission rate E, the room volume VR and the given air change rate λ, whereas the points representtheconcentrations that have been measured (bedroom: triangles; office space: squares). It is evident that the measurements in the bedroom oscillate about values which would be expected for an ACR of about 0.05 h-1. The reason for the large deviation of the two first measurements from the curve is probably due to largely differing flow resistances of the Tenax tubes used. In comparison to the bedroom, the CO2 values in the office (lower curve) show a much lower level despite the double emission rate. On the one hand, the lower concentration increase is caused by the larger room volume, but another reason is the significantly higher ACR of 0.45 h-1. The calculation of the ACR from the individual concentration values measured led to values ranging from 0.05 to 0.14 h-1 for the bedroom and from 0.2 to 1.7 h-1 for the office space, respectively.

Figure 7.

Course of CO2 concentration in a bedroom equipped with an insulating glasswindow with rubber sealing. Window and door closed. Measured values (black) with corresponding air change rates [in ACH], dotted curve: theoretical values. Room volume:24 m³, occupancy: one adult.

Figure 8.

Course of hexafluorobenzene concentration (HFB) in two rooms with different volumes and different window types, theoretical curves calculated from given emission rates E, air change rates λ, known room volumes VR. Black glyphs: measured values.

The results of the tracer gas comparison are shown in Table 5 for both test series. At large, a very good agreement was found for ACR determined with the two different tracer gases.

Table 5.

ACR results of comparative measurements with CO2 and hexafluorobenzene as tracer gases. Windows and doors closed.

3.4.2. Parallel testing of CO2 and SF6

In another series of investigations an office space with a volume of 48 m³ and two double box windows was doped with CO2 and SF6 in parallel on 6 consecutive days, and the ACR were determined from the decay curves. Measurement of SF6 was done in the same way as described in the above section "decay curve". An example of the parallel concentration decay of both tracer gases is given in Figure 9; results of the 6 experiments, which differ only slightly, are compiled in Table 6.

Figure 9.

Decay curves of CO2 and sulphur hexafluoride (SF6) and curve fits; simultaneous measurements in the same room.

Table 6.

Comparative measurements with CO2 und SF6; office space (volume 48 m³, two double box windows); COV (%): coefficient of variation = 100*standard deviation (std.dev.) / average.

The average ACR was 0.304 h-1 for CO2 (SD ± 0.09 h-1) and 0.298 h-1 for SF6 (SD ± 0.09 h-1). To check for systematic differences between the individual determinations we calculated the ratio of the two air change rates and got a mean ratio of 1.02 with a standard deviation of ± 0.08. Thus, no statistically significant systematic differences between the compared tracer gases could be detected. Since our sample size was rather small (only 6 determinations for each tracer gas) we included results from other studies (Guo and Lewis 2007; Roulet and Foradini 2002; Shaw, 1984) for comparison. After inclusionof these data the difference wasstatisticallysignificant, albeit the systematicdeviationbetweentheACR, determined under the same conditions with either CO2 or SF6, was very small (Fig. 10). A similarly good agreement was found by Stavova (Baránková) et al. (2006) comparing CO2 and Freon 134a.

The mean ratio of the two air change rates (CO2/SF6) is about 1.06 (95% CI: 1.01 – 1.12). Standard deviation ranged from 0.93 to 1.21. The ≈6% overestimation of the ACR, as obtainedwithCO2, is yet much smaller thanthe 50percentoverestimation obtainedby Schulze and Schuschke (1990) who used krypton 85. On the basis of existent data it is so far not possible to conclude with certainty whether or not very small ACR (i.e. 0.1 h-1) are responsible for higher systematic deviations.

3.5. Wind and temperature effects on the natural air flow in case of joint ventilation

Natural ventilation is driven by the air-pressure difference between theinteriorandexternalenvironment which prevails on the room's openingareas like windowsanddoors. The air change rate is also influenced by thetightnessof windowsanddoors versus jointsand cracksin the building envelope. The air-pressuredifferenceis predominantly causedby temperaturedifferencesbetweenindoorand outdoor climate,andbylocalwindeffects. Onthewindward sideof a building, the wind is trapped reaching its highestdynamic power when the flow angle equals 90 . Onthesides of a building that parallel thewinddirectionandon the far side, negative air-pressures occur regularly. These air-pressuredifferencesresult in raised rim hole rates, particularly in case of leaksor openwindowson opposite sides of a building, due to suctionand pressure cycles of the wind(i.e. cross ventilation). The entire air-pressureontothefaçadeisdependentonboth thewindspeed and squalls,as well as the building’s constructional quality, namely the tightness of thefaçade.

Figure 10.

Comparative air change determinations with CO2 and SF6 as tracer gases. Results from other authors and own results. Solid line: mean, dotted lines: 95% confidence interval of the mean, dashed lines: interval of ± 1 standard deviation.

3.5.1. Wind influences

Tostudy theeffectsof weather conditions on the ACR,measurementswere conducted intwo different officebuildingsof theRobertKoch Institute, Berlin, Germany. Building #1 is anolder brick building (built ca. 1900), surrounded by allotment gardens (to the northandsouth) and largeopen spaces. On the east side is a three-storey row house, and on the west side afour-storey office building. Dueto the location of the building, unhindered wind flow is limited to south-east or south-west directions. All of the five rooms which have beenselected in building #1 were located on the south side and equipped with double box windows (rooms A– D and F; room volumes between 30and 90 m³).

The rooms in building #2 constructed in the late 1970s were equipped with tightly closing windows featuring single glazing and insulating rubber seals. The room selected for the measurements had a large window sill with a couple of windows in north-east direction and a room volume of 40 m3 (room E). In front of the windows was an open space (length ≈150 m), limiting wind flow to solely from north-easttosouth-east directions. On the south side,room E wasprotectedbya cross-building located in a distance of ≈10 m. Likewise, on the north side, the building was protected by a north-east to south-west oriented wing of the same building (in a distance of about 70 m from room E).

Between January 1999 and December 2010 a total of 611 air change measurements were performed in the five rooms of building #1 (A – D and F) using the CO2 decay method. Thedecay curveswere mainly recorded at night or on weekends according to ASTMD6245-98,when the building was unoccupied. Analogously, 388 air change measurements were carried outin roomEof building #2 between July 2002 and September 2008. During the measurements all doors and windows were kept closed. Data about meteorologicalconditions during thetests, such as outsideairtemperature, wind speed andwinddirectionwereobtainedfrom the web sites of local Berlin weatherstations. In roomEofbuilding #2 the indoor temperature was recorded during the measurements. Table7provides an overview ofthe air change rates determined. Forroomswithdouble box windows individual values ofACR varied from <0.05 h-1 to 1.8 h-1(i.e. a factor of ≈40), whereas the medians (50th percentile) obtained for the rooms A – D and F varied only between 0.3 and 0.6h-1. Significantdifferencesbetween the individual roomscouldnot be detected.

Table 7.

Air change rates [ACH] in office spaces with windows of different construction, rooms A-D, F: wooden double box windows, room E: insulating glass window with plastic frame and rubber sealing

Significantly lower ACR were determined for room E in building #2 equipped with double glazed windows for which the median was 0.08h-1. The individual values varied between 0.016 and 0.40h-1, i.e. maximum and minimum differ more than a magnitude. Wind influences, especially wind speed, were identified as the main reason for the variability of theACRin the two buildings (Fig. 11).

In building #1 (rooms A – D and F) theACRrangedfrom<0.1 h-1 when windspeed waslower than 5 km/h to1.2–1.4h-1at a wind speed up to 30–40 km/h. The strong influenceofthewind caused significantanduncontrollableheatlossesthrough leaky windows which, in turn, caused unhealthy draught effects.

Surprisingly, the results obtained for room E in building #2 equipped with insulated windows and rubber seals, were significantly dependent on wind influences, as well, although the wind effects weremuchsmallerthan those of building #1. For room E,ACR were less or equal to 0.05h-1 when wind speed was lower than <5 km/h but increased up to 0.25h-1with a wind speed ofmorethan 40 km/h. Therelations describedhereessentially confirm the results formerly obtained by Wegner (1983). Besides the wind speed, theflow directionof the wind had a significant impact onthe amountof the ACR. Thisisevidentfrom the graphical presentation in Figure12, grouping ACRs according to the winddirection prevailing in test periods.

Figure 11.

Influence of wind speed on ACR in office spaces equipped with windows of different construction. Windows and doors closed. Rooms with double box windows (A-D, F: n= 611), room with insulating glass window: n= 388).

Figure 12.

Wind direction and ACR in office spaces equipped with windows of different construction. Windows and doors closed. Rooms with double box windows (A-D, F: n= 601), room with insulating glass window: n= 388).

Adverse wind blowing against the window or in parallel to the window front are of higher impact than wind flow from the backside of the building, the latter causing lower ACR than the two former (see Table 8).

Table 8.

Dependance of ACR on wind direction, raw and adjusted ACR quotients, Rooms A-D, F: windows to south, room E windows to north-east.

In case of the roomswithdouble box windows, this led toarelative threefold increase ofairchange, expressed as ratio ac(luv) : ac(lee), for wind directions from south towest. In theroomwithdouble glazed windows, we measured an almost doubled increase ofairchange (70– 100%)for winddirectionsfrom north toeast. The overall airchangeratios depending upon the relative wind direction, respectively were not remarkably influenced by the wind speed (Quotientadj in Table8). Summarising wind effects we could demonstrate thatregardless ofwind strength and the windowtypeanairflow directed tothe window front can yield a 1.5- to more than 2-fold increase of the ACR in comparison to wind from the backside.

3.5.2. Temperature influences

The temperature difference between inside and outside is one of the major forces driving ventilation. We investigated the impact of the temperature gradient (room air temperature minus outside air temperature) on the ACR in room E and found astatisticallysignificant effect on the extent ofairchange(Fig. 14a). Whereas the average air change rate λ, after correction for the wind influence, was only 0.06 h-1 at atemperaturedifference ΔT of 0K, it increased up to 0.1 h-1 when the temperature gradientrose toabout30 K. However, as compared to wind effects (Fig 14b), the exponential relationship between ACR and temperature gradient(Fig.14a), turned out to be much weaker, at least for room E with its tightly closingwindow.

3.6. Air change rate during window ventilation

Wind action and temperature gradient both proved to be significantfactors influencingthe amountofairchange through joints and cracks in buildings. For indoor spaces in buildings with high leak tightness, windowventilationplaysacrucialrole in the removalof excess humidity, carbon dioxide, odorsand othervolatilecompounds. Theeffectiveness ofventilationviaopenedwindowsisnot only dependent ontheir number, sizeandarrangementin the room, but also on the modus and frequency of active window ventilation. Windows that are only opened partway (i.e. tilted) ventilateless effectivethan fully openedwindows. Cross-ventilation can be achieved when windows on oppositesides ofa room are opened simultaneously. Cross-ventilation removes moistureandcontaminants most effectively fromthe indoor airtothe outside. Theeffectiveness ofthedifferentmodes of window ventilation expressed in terms of the ACR, was examined with a couple of tests in room E, the results of which are shown in Figure 13 in conjunction with thejointventilation results.

Figure 13.

ACR and window position. Office room E.

ACR determinations were carried out by the decay method; duration of themeasurements varied between 0.5h and3h with recording intervals of 12 sto 1 min.After doping the room with CO2, one window in the room (dimensions: width 0.8m and length 1.2m) was either in tilted position(i.e. 10 cm wide upperopening)orwidely open(i.e. opening angle80 ). To achieve cross-ventilation, an opposite window was also widely open. For both ventilation modes – tilted or full open – the following ACR were determined (Table 9a): With the tiltedwindow a total of 239 measurements was performed; the average ACR(i.e. median) was 1.5 h-1 with day-to-day variations that ranged from 0.5 h-1to3.4h-1. Inthe caseof full open window, the median ACR was10h-1, with a minimum of 2.4h-1 and a maximum of 30h-1 (321 measurements). For cross ventilation a number of 179measurementswere performed. The ACRvaried between 15 h-1 and 146h-1(median: 40h-1). On the basis of these data we calculated the ventilation times which are necessary to reduce an indoor CO2 concentration of 1500 ppm by 98% (corresponding to a CO2 concentration that is about 30 ppm higher than outside) and yielded 2 hours for the tilted window (range 1 – 7 hours), ≈0.3 hours for the wide open window (range 0.1 – 1.5 hours) and 0.1 hours for cross-ventilation (range 1 - 14 min; see also Table 9b).

Table 9a.

Dependance of ACR on window construction and opening

Table 9b.

Ventilation time needed to reduce CO2 concentration of 1500 ppm by 98%

These results clearly confirm the great importance of ventilation with fully open windows and, even more, of cross-ventilation for the efficient removal of indoor pollutants. Thebroad variation inACR can be explained to a great deal by theeffectof the temperature gradientbetween indoor and outside (Fig. 14a).

Compared to the temperature gradient,windeffectsplayed only a minor role at least for the window openings "tilted" and "widely open" tested here (Fig. 14b). The dependence oftheACRonthetemperaturegradientcouldbe estimated by non-linear exponential curve fitting for all tested window opening modes (Fig. 14a). From these findings we can conclude thatin case of ventilation via tilted window, witha temperaturegradient of10K,an ACR of0.92*exp(0.045 * 10) =1.4h-1 can be expected for the room examined here (95% prediction interval:1.0h-1–2.0h-1). With atemperaturegradient of20Kan average ACRof2.3h-1 can be expected (95% prediction interval1.6h-1 – 3.1 h-1). For ventilation with fully open window an ACR of 4.5 * exp (0.071 * 10) = 9.2 h-1 (Δ T = 10 K) can be expected and 18.6 to h-1 (Δ T = 20 K) with 95% prediction intervals 5.5 h-1– 15.3 h-1 and 11.1 h-1– 31.2 h-1, respectively. In the case of cross ventilation an average ACR would expected to be 39.0 h-1 and 50.2 h-1 for Δ T = 10 K and 20 K, with 95% prediction intervals of 23.0 h-1 - 65.8 h-1 and 29.6 h-1 – 85.1 h-1, respectively.

3.7. Air change rates in Berlin's housing stock

Theresultspresented so fardo not allowgeneralstatementson theairchangeinbuildings, sincethenumberofrooms examinedwas far toosmall. As partofsite inspections and indoor air investigationsduring 1999–2005,ACR were determined for a further number of different buildings and rooms applying the CO2 concentration decay method. A total of 198 livingandworkingspaces, situated in 152 buildings in theBerlin areaandin the immediatesurrounding werestudied, comprising of 143residentialbuildingsof different

Figure 14.

a. Influence of temperature gradient (indoor to outdoor) on ACR, corrected for wind effects, b. Influence of wind speed on ACR, corrected for temperature effects, in an office space (room E) for different ventilation conditions: natural ventilation: "closed", "window tilted", "widely open" and "cross ventilation", (from bottom to top); room volume 40 m³, solid lines: fitted curve, dashed lines: 95% prediction intervals.

age and storey numbers, and nine different office buildings. None of the rooms had an additionally installed ventilation system. Tenants were asked not to enter the rooms duringthetracergasmeasurements and to keep windows and doors closed. The duration ofthetracergasmeasurementsvariedbetween2 hoursand 24hours. To obtain decay curves with a high number of nodes we choose time intervals of 1 to 3 minutes between CO2 measurements. Two Testo 400 devices withCO2probe(infrared absorption) were used for recording and data logging. The results show that under worst case conditions the vast majority of the ACR are below the range of 0.5 h-1 – 1 h-1 which was recommended by the former German Federal Health Office (Bundesgesundheitsamt, 1993) for common living quarters. The ACRof the studiedrooms ranged from minimum0.02 h-1to maximum1.98 h-1. Of the198naturallyventilatedrooms the majority (167=84%) had ACRof 0.5h-1orbelow. ACR of 0.8h-1or higher could be determinedfor only 12of the rooms (6%). The50th percentile(median) of the ACRwas0.2h-1witha95% confidence interval(CI) of 0.17 h-1 – 0.24h-1; arithmetic mean was 0.31h-1 (95% CI 0.27 h-1 – 0.36h-1) and geometric mean was 0.22h-1 (95% CI 0,20 h-1 – 0,25 h-1), respectively (see Table 10). Thus, our studies essentially confirm the results obtained by Münzenberg (2004) and Salthammer etal. (1995), who conducted ACR measurements under similarmethodologicalconditions inGermany, and received results consistent with those from surveys in Scandinavia by Andersen et al. (1997); Bornehag et al. (2005) and Harving et al. (1992).

Measurements performed in the same rooms using diffusion tubes and hexafluorobenzene (C6F6) or PFT as a tracer gas yielded somewhat higher ACR compared to that obtained with the CO2 concentration decay method. This can be explained by necessarily longer exposure times of collection tubes ranging from several days to weeks, which essentially means that there have been one ore more active ventilation phases during the measurement period influencing the air change (Table 10). However, eveninthesestudies, the proportion of roomswithACR below 0.5 h-1was remarkably high. Studies conducted in the USA also show a high proportion of buildings (50 percent and more) with ACR below 0.5 h-1 (Persily et al., 2010), suggestingthatabasically desirable ACRof at least 0.5h-1is difficult to achieveunder normalconditions. Similar conclusions have been drawn by Erhorn and Gertis (1986) and Münzenberg (2004).

Stratification for window types showed significantly higher ACR (factor ≈2) for rooms with double box windows or composite windows than for rooms with specially insulated windows (double glazing and continuous rubber seal; Table 11). However, also for rooms with double box windows or composite windows the majority of the ACR determinations yielded values <0.5 h-1. The50th percentile(median) was 0.33h-1, and the75th percentilewasabout0.56h-1. In comparison, rooms with double glazed windows had a median ACR of 0.16 h-1 and the 75th percentile was 0.25 h-1.

Table 10.

ACR measurements in Germany and Scandinavia, results of surveys. TFE: 1.1.2-trichloro-1.2.2-trifluoroethane, 1) before reconstruction. 2)after reconstruction, 3) single family houses, 4) multi-family houses, AM: arithmetic mean,. GM: geometric mean, n.a. not available, < 0.5 h- 1: proportion (%) of air change rates < 0.5 h-1

Table 11.

ACR in naturally ventilated rooms (windows and doors closed) equipped with different window types. Min: Minimum. Max: Maximum. 05. P: 5th percentile. 25. P: 25th percentile. 75.P: 75th percentile. 95. P: 95th percentile. 95% CI: 95% confidence interval. AM: arithmetic mean. GM: geometric mean

Theempiricalfrequency distribution of the ACR for thetotalsamplehadthe typicalshapeof alog-normal distributionwith asteeprise onthe leftand a long tail on therightside (Fig. 15a). A good fit to the normal distribution could be achieved by calculating the logarithms of the ACR (Fig. 15b).

Figure 15.

Empirical frequency distribution of ACR (natural ventilation) for 198 rooms in the City of Berlin; a, values as determined; b, log-transformed values. Solid line represents normal distribution.

This is consistent with the results of Scandinavian studies by Andersen et al. (1997); Bekö et al. (2010); Harving et al. (1992); Øie et al. (1998) and Ruotsalainen et al. (1992) who found the same form of empirical frequency distributions of ACR in Danish, Finnish and Norwegian residential buildings.

3.8. Ventilation and pollutant concentrations

Adequate ventilation is the cheapest and easiest way to effectively remove indoor pollutants. Previous studies have shown that insufficient airchangeis mainly due to renovation and other structuralchanges, namely in most cases the installationof new air-tight windows (Wegner 1983, 1984). The results of our own investigations presented here have led tosimilar conclusions, apart from significantly lower ACR as compared to previous studies(Wegner 1983, 1984). Generallyinsufficientfresh air supply resulted in anconsiderableincrease of indoor airpollutants. This is the caseforformaldehyde(Li et al., 2002; Salthammeretal., 1995; Wegner,1983) and othervolatilecompounds(Hodgson etal., 2003) as well asforthe radioactivenoblegasradon(Andersen etal., 1997; Chaoetal., 1997; Sentikova, 1999). Our own studiesalsofounda correlationbetween theconcentrationof volatileorganiccompoundsand the ventilation rateinnaturallyventilatedrooms, as described below.

We have chosen some frequently found indoor air pollutants, which may be taken as typical representatives of different substance classes, namely formaldehyde, acetaldehyde, acetone, 2-ethyl-1-hexanol, hexanal, d-limonene, α-pinene, decamethylcyclopentasiloxane (D5-siloxane), and toluene. Air sampling to analyse thesecontaminants was done in the rooms under study (closed windowsanddoors), either after a CO2tracergasmeasurement was finished or directly before the start of a CO2 decay recording with Aircheck samplers (model 224-PCXR8) SKC Inc., Eighty Four, PA, USA. Forenrichmentweused the adsorption media recommended for these pollutants. Airvolumes varied from 90 l to 120 l and the sampling airflow was 1.5– 2 l/min. Formaldehyde was somewhat special since itwasmeasured semi-quantitatively in situ inmostrooms. An Interscan 4000 device (Ansyco, Lenzkirch, Germany) was used to carry out the measurements. Prior to the determination of the formaldehyde concentration, a zerobalance was produced bypumpingairthroughan upstreamactivatedcarbonfilter for 10 minutes. Thereafter, the concentrationvaluewas recordedwhenthe readingremainedconstant. Comparison ofmeasurementsfor24different roomsyieldedagood agreement of concentration levels obtained with the electrochemicalformaldehyde method using theInterscan4000and the laboratory results obtained by analysis of the airsamplesthat were collected with DNPHcartridges.

Statistics for the pollutants investigated here are listed in Table12; if the sample size was sufficient the95% confidence intervals were also included. Acetone, formaldehydeandacetaldehydeshowedthe highestconcentrations, with median valuesof75µg/m³, 23 µg/m³and 31 µg/m³, respectively. Due toboiling pointsbelow60 C, these pollutants are classified as very volatileorganiccompounds (VVOC). Therefore, they are–withthe exceptionofformaldehyde–notregularlyrecorded indoors.

As Figure16shows, thefrequencyofformaldehydeconcentrations >100µg/m³ seems to correlate withthe ACR. In8of77 category-1 roomswithACR<0.2 h-1formaldehyde amounted to ≥100µg/m³, i.e. the frequencyof exceeding values was 10.4% (95% CI 4.6% –19.5%). Among the category-2 und -3 rooms was only one each with exceeding formaldehyde (1.6% and3.2% relativefrequency, 95% CI 0% –8.5% and0% –16.7%, respectively). According to Χ2testing these differences were close to be significant (p=0.060), but this seems unlikely due to strongoverlapsof the very broad 95% CI. Our finding thus points to successfulemission reduction measures through low-emission products by which pollutant concentrations can be accomplished that are within the range of the accepted limits, even in very airtight rooms.

Figure 16.

Frequency of formaldehyde concentrations >100 µg/m³ and ACR (natural ventilation) in rooms of the City of Berlin.

Figure 17.

a. and b. Concentrations of acetaldehyde (a) and acetone (b) and corresponding air change rates in rooms of the City of Berlin. Sampling: worst case conditions, solid line: mean concentration, dotted line: 90% prediction limit.

Unlike formaldehyde, the concentrations of acetaldehyde and acetoneshoweda cleardependenceofthe airchangerate λ, which could be described by the power function (C(λ) = a*λ-k). The independent variable air change rate could explain about 50% (R2 = 0.5) of the total variance of the acetaldehyde concentration values and about34% (R2=0.34) of the acetone concentration values. The results of a curve fitting are shown in Fig. 17 a and b.

According to the resulting functionparameters it can beestimatedthat an increase ofthe ACR from 0.1h-1to 0.5h-1 would yield a reductionof the indoor concentrationsfrom 40µg/m³ to 15µg/m³foracetaldehyde,and from 145µg/m³to60µg/m³for acetone. This is a reduction factor of 2.4 – 2.7.

Table 12.

Concentrations [µg/m³] and 95% confidence intervals of selected VVOC and VOC in the air of rooms in the City of Berlin; n.d.: 95% confidence interval not detected (small sample size), 10. P: 10th percentile, 90. P: 90th percentile, AM: arithmetic mean, GM: geometric mean.

Similar relations were found forothervolatileorganiccompounds(Table 12). Again, thedependenceof their indoor airconcentrationsonthe ACR could be characterizedby a powerfunction (see Figures 17c to 17h). Regression analysis revealed that an increase of the ACR from 0.1h-1to 0.5h-1 decreased the concentrations of 2-ethyl-1-hexanol andhexanal (or hexanaldehyde) by a factor of ≈2.3, since 2-ethyl-hexanol declined from 6.2µg/m³ to 2.6µg/m³, and hexanal declined from 21.2µg/m³ to 9.1µg/m³. Forlimoneneandα-pinene the reductionfactorswere2.6and2.2, respectively. For these pollutants, the indoor air concentrations would be reduced from 24 µg/m³ and 30 µg/ m³ to values of 9 µg/m³ and 14 µg/m³, respectively For decamethylcyclopentasiloxane(D5) and toluene the reduction factors amounted to2.7and2.4, corresponding to a decline 6.5µg/m³to 2.4µg/m³(D5) or32µg/m³to 13µg/m³(toluene).

Figure 17.

c-h.Concentrations of selected volatile organic compounds (VOC) and corresponding ACR in rooms of the City of Berlin. Sampling under worst case conditions, solid line: mean concentration, dotted line: 90% prediction limit.

These results may allow to conclude thatincreasingthe ACR to the fourfold will reduce indoor airconcentration of VVOC and VOC to half. Thismeansthat theconcentration is inversely proportionalto thesquare rootof theACR, but seems to be contradictory to theoreticalconsiderations, ifoneassumesanearlyconstantemissionrateof volatilecompounds. In this case, the concentration would be inversely proportionalto theACR underequilibriumconditions, meaning that a fourfold increase in the ACR would reduce the concentration to a quarter.

One of the reasons for the deviating results is surely the fact that nowadays many buildings are constructed with such a high air-tightness that a kinetic equilibrium cannot be achieved under typical sampling conditions (i.e. time span of several hours between last window ventilation and start of the measurement; see also Figure 3b). Thatleads, firstly, tolowACR (<0.2 h-1) and, secondly,to an underestimationof the VOCconcentrations. Theregressionanalysis yieldsaflattercurveand anexponentk<1 forthe air change rate λ. Anotherreasonis thedifferentinterior decoration (furniture, wallpapers, other equipment) of the spaces investigated here, andthustheir differing pollutionloads. Our VOC sample comprised rooms which differed greatly in respectto number and kind of possiblesources. Therefore, we can assumethateven in roomswith lowairchange either no VOC sources existed or only such sources withlowemissionrates. As a consequence,onlylowconcentrations could have been measuredin the air of theserooms. This fact would also contribute to the flattening of the air change-concentration curve.

Theinfluence ofACRonthe concentrationsof VOC is also valid when we consider substance classes. The spectrumofindividualcompoundsof these classes resembles that described by Schleibingeretal. (2001, 2002). Aswithformaldehyde, the analysis wasperformedwithgroupeddatato revealthedependency onthe ACR, and categorisation was done as above. For determiningoffrequencies of exceeding concentrations the individual measurement values were summed up. Subsequently, the sum values were dichotomised on the basis of the medians of the empirical distributions. The empirical medians agreed quite well with the limits proposed by Schleibinger et al. (2002) – especially when taking into accountthe 95% confidence intervals of the percentiles (Table 13). Exceptionshere are thesiloxanes, for which the 90th percentile was clearly above Schleibinger's recommendation (Schleibiger, 2002). The marked deviation from previous studies (Schleibinger et al., 2001; Scholz, 1998) may indicate an increased (and still increasing) use of siloxanes as constituents of coating materials, paints, cosmetics etc., thus contributing more than ever to indoor air pollution.

Table 13.

Indoor air concentrations and 95% confidence intervals [µg/m³] for selected classes of chemical compounds. Results of a survey performed in the city of Berlin. Published target and reference values by Schleibinger et al. (2002) are included. 90. P: 90th percentile, AM: arithmetic mean, GM: geometric mean.

The relationshipbetweenambient airconcentrations of selectedsubstanceclassesandtheACR the following picture can summarised as follows (Fig. 18): Dependingon the substanceclass,between 50% and 70% of roomswithACR<0.2 h-1(sample sizeof this subgroup varied from n= 41ton= 53) showed concentrations whichwereabovethemedianof the respective substance class. For ACRbetween0.2h-1 and0.5h-1(sample size between n =37andn= 52) this was true for only 40 – 50% of the rooms. This proportion was further reduced to 10 – 20% when the ACR were equal to or greater than 0.5 h-1 (n = 15-27). This trend was statistically significant for all substance classes studied here (χ2 test for trend: aldehydes p <0.01, aromatics p <0.001, terpenes p <0.001), Even for siloxanes the trend was significant, although not very pronounced (p <0.05).

Figure 18.

Frequency of concentrations above median (50th percentile), given for selected compound classes and stratified by ACR. Active air sampling under worst case conditions, error bars: 95% confidence intervals.

Overall, it canbe deducedthattheprobabilityofexceeding pollutant concentrations not only depends on the numberofindoorsources andtheir emitting power, but alsoon theair-tightnessof rooms. However, no exceeding concentrations can occur without the presence ofindoorsources. Forthisreason, theuseof low-emissionproducts should givenpriority in order to avoid preventable future exposure to air pollutants. Whether a given indoor air concentration is above the usual burden and appropriate remedial measures are necessary, can be clearly evaluated by theACR.It would therefore bedesirabletodefineair-change-related referencevaluesfor common and tolerable concentrationsof volatileindoorairpollutants.

4. Conclusion

Tracergasmeasurementsfacilitate the determinationofairchangebetweenindoor and outdoor environmentunder different ventilation conditions. Inthis papermethodsare described which can be used to determine the air change rate with tracergas measurements. Special emphasis was given tocarbon dioxide, since this tracergasisoftenusedto assess indoor airqualityandairtightness ofrooms. Wehaveshownways todetermine airchangerates with carbon dioxide and compared the results with those obtained by other tracergases.

The dependence ofthenaturalair change fromtheprevailingweatherconditions, such as thecurrentwindandtemperatureconditionscouldbedemonstrated by owninvestigations. Wheather effects in conjunction with different window types were leading to considerable day-to-day variability, thereby limiting the reproducibility of measurementsto a large extent. However, focusing on therelationship between air changeand window ventilation we could showthatventilationwithwide openwindows or cross-ventilation are by far thecheapest ways to remove indoor pollutants as quickly as possible from indoor air. The effectiveness of window ventilation strongly depends on the temperature difference between room air and outside air, whereas the influence of wind seems to be less important.

Studieson theair quality inBerlin'shousing stockrevealeda very highproportionofrooms with lowair changerates. About80% of therooms examined by usshowedair changeratesbelow 0.5 h-1, half of which were even below0.2 h-1, when determined under so-called "worst case" conditions. Especiallyin roomsthathadnewlyinstalledwindowswithinsulated glassandperipheralsealingstrip, verylowair changerateswere frequently determined. Of these,about50% (median) had air changeratesbelow 0.16 h-1. This was in significant contrast to rooms in older buildings equipped with wooden framed double box windows or composite windows, which displayed mean air change ratesof0.47h-1.

Lowair changerates often contributetoelevated VOC concentrations, in rooms with insulated windows, unless emissions are notlimited byappropriate low-emissionproducts. This relationshipcould beconfirmedby our investigations. Improvementof ventilation i.e. increasingtheairchangerateis an efficient measure to additionally reducethepollutantloadin indoor spaces. Therefore,to achieve limitationofunhealthy indoor-borne pollutants a combination of both measurescan strongly be recommended.

5. Acknowledgement

We are grateful to Dipl. Ing. P. Tappler, Innenraum Mess- und Beratungsservice, Vienna, Federal Republic of Austria, for providing instrumentation and sulphur hexafluoride for the measurement of air change rates. We also thank Dr.-Ing. H. Schleibinger, Institute for Research in Construction, National Research Council of Canada, Ottawa, for his generous assistance in the acquisition of literature, and to the ALAB staff, Analytic Laboratory Berlin, for carefully conducting the sample analyses. We gratefully thank Dr.Liv Bode for her support in revising the manuscript.

© 2011 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike-3.0 License, which permits use, distribution and reproduction for non-commercial purposes, provided the original is properly cited and derivative works building on this content are distributed under the same license.

How to cite and reference

Link to this chapter Copy to clipboard

Cite this chapter Copy to clipboard

Detlef Laussmann and Dieter Helm (July 27th 2011). Air Change Measurements Using Tracer Gases: Methods and Results. Significance of air change for indoor air quality, Chemistry, Emission Control, Radioactive Pollution and Indoor Air Quality, Nicolas Mazzeo, IntechOpen, DOI: 10.5772/18600. Available from:

chapter statistics

3021total chapter downloads

5Crossref citations

More statistics for editors and authors

Login to your personal dashboard for more detailed statistics on your publications.

Access personal reporting

Related Content

This Book

Chemistry, Emission Control, Radioactive Pollution and Indoor Air Quality

Edited by Nicolas Mazzeo

Next chapter

Olfactory Comfort Assurance in Buildings

By Ioan Sarbu and Călin Sebarchievici

Related Book

First chapter

Planning Air Pollution Monitoring Networks in Industrial Areas by Means of Rremote Sensed Images and GIS Techniques

By Mauro Rotatori, Rosamaria Salvatori and Roberto Salzano

We are IntechOpen, the world's leading publisher of Open Access books. Built by scientists, for scientists. Our readership spans scientists, professors, researchers, librarians, and students, as well as business professionals. We share our knowledge and peer-reveiwed research papers with libraries, scientific and engineering societies, and also work with corporate R&D departments and government entities.

More about us