Sterling Vineyards stores barrels of wine in both an air-conditioned, unheated, unhumidified building (chai) and a cave (capacity: 4,000 barrels), in which there are no mechanical means to affect the environment.

An experiment was designed to quantify differences in topping wine volume and costs between the chai and the cave. Twenty barrels containing 1988 Sterling Reserve (a red Bordeaux variety blend) were stored in each location and monitored for 50 weeks.

The barrels employed were 225-L, 1988 Seguin Moreau, Allier forest origin, château ferré style, medium + toast, and approximately 20-mm thick staves. The barrels had been used previously for Chardonnay fermentation and ageing from September 1988 until April 1989, when they were filled with the Reserve blend.

In order to study the effects of environmental factors on evaporative loss from barrels, temperature and relative humidity (sling psychrometer) were measured weekly, and the barrels were topped every five weeks with a graduated cylinder.

White Livingston atmometers were utilized to study the evaporative power of the air. These atmometers consist of a hollow porous porcelain sphere 5-cm in diameter, which is filled with water and mounted on a water reservoir bottle. Since the nite of water loss from the porcelain surface is affected by air move- \1 ment, humidity and temperature, the true evaporative power of the air can be measured.

Relative humidity in the chai showed large weekly fluctuations, presumably due to both atmospheric humidity changes and dehumidification by the air conditioning.

Dehumidification is a function of both the amount of time the air conditioner operates and the glycol temperature in the air conditioner. The mean temperatures and relative humidities for the chai and cave were 13.5°C (56.3 °F) / 73.8% and 16.7°C (62.1 °F) / 92.5%, respectively. Atmometers, placed in both sheltered and exposed locations from mid-head to bung-high among bottom-row barrels, revealed only very small differences in evaporativity within a group of barrels.

Temperature variations from one topping to the next cause expansion or contraction of the wine, which can result in major aberrations in the apparent evaporative wine losses. All topping volumes, therefore, were corrected for expansion 1 contraction of the wine.

The volume coefficient of expansion of the wine was determined by weighing a wine-filled volumetric flask at several temperatures. A typical value of the volume coefficient of expansion of the wine, which is a function of temperature, is 0.024°C at 15°C (O.013 °F at 59°F). This corresponds to 54 mL/°C (30mL/°F for a 225-L barrel). Expansion 1 contraction of the barrel itself was shown to be negligible.

Fig. 3 shows the topping requirements, corrected for expansion 1 contraction, for all five-.week toppings. Each bar in Fig. 3 represents the mean volume needed for each 20-barrel lot.

At each topping, significantly more wine was required to top barrels in the chai. This observation was confirmed by t-tests (all p < 0.001) . The ratio of the mean topping requirement for the entire 50-week period for the chai to that for the cave was 2.61.

The mean topping wine used was 6.82- L/barrel per year (3.03) in the chai and 2.61-L/barrel per year (1.16) in the cave. The difference between these percentages can be used to determine the monetary savings resulting from lower wine loss in the cave.

For instance, this difference amounts to nearly 1/2-case of wine per barrel per year, which translates to about $1,700 per 1,000 cases at a wholesale price of $7.501 bottle.

Two-way analyses of variance of the topping requirement data, corrected for expansion / contraction, for each location show significant variations (p < 0.001) for both topping means and barrel means. This demonstrates both the effect of seasonal environmental changes and barrel-to-barrel variation.

Fig. 4 shows the frequency distributions of the mean barrel topping requirements. The need for a large sample size for barrel experiments is evident from the large variability observed.

Although evaporative losses of wine do correlate to some extent with relative humidity alone, evaporation is also a function of temperature, air speed, and other factors. While the effect of many of the factors can be very complex, the evaporation rate of water should be proportional to its vapor pressure deficit, which is a function of relative humidity and indirectly of temperature.

When water and air temperatures are equal and surface and boundary effects are negligible:

Vapor Pressure Deficit=px[1-(RH%/100)] where p is the vapor pressure of water at ambient temperature. Since no appreciable concentration of ethanol exists in the air, the vapor pressure deficit for pure ethanol is simply the vapor pressure of ethanol.

An estimate of the volume of ethanol that evaporated during each topping interval can be generated from overall topping needs, composition changes as determined by ethanol analyses, mean temperatures for the topping intervals, and vapor pressures of ethanol at those temperatures.

If the ethanol volumes are subtracted from the total volumes, water evaporation can be determined. Fig. 5 shows water evaporation as a function of water vapor pressure deficit.

Three outliers are evident. The excessive evaporation for the two at higher vapor pressure deficit, which correspond to the chai in the middle of the summer, can be explained by increased evaporation caused by significant air movement from the nearly constantly operating air conditioner. The least squares line for all data except these two mid-summer outliers (R= +0.96) is shown.

The fact that water loss correlates well with the water vapor pressure deficit allows the evaporative loss from barrels to be estimated at various relative humidities and temperatures (Table I).

Estimated alcohol losses have been added in to give total wine loss per annum. The data only applies to conditions similar to Sterling's: table wines stored in tight-grained, 225-L, château (thin-stave) barrels in non-windy areas, etc.

Even under quite different conditions, Table I should give a good idea of the effects of the environmental parameters. For example, the evaporation rate at 16.0°C (60.8°F) 155% RH should be about twice that at 13SC (56.3°F) / 75% RH.

Atmometers, which were placed in the chai and cave at mid-head height between bottom-row barrels, were weighed at the beginning and end of two topping intervals. The rate of water evaporation per tJ unit surface area was much greater for atmometers than for barrels. The ratio of loss per unit surface area ranged from 36 in the cave, for a period when doors were always closed to 105 in the chai for a period in which the air conditioner was operating often.

This leads to the conclusion that the loss from barrels is slowed greatly by the need for the wine to migrate through the wood. The variation in ratios indicates that evaporation from the atmometer surface is more sensitive to air movement than barrel surfaces.

The above data implies that the relative humidity balance point at which a table wine doesn't change in ethanol concentration due to evaporation is greater than 74%, the overall mean for the relative humidity in the chai. This inference is consistent with some data from J.F Guymon, but not with the 60 to 65% level reported elsewhere.

Atmometry showed that migration of wine through wood is slow compared to evaporation. The relative humidity balance point at which wine does not change in ethanol content appears to be greater than 74%. Above this balance point, ethanol concentration will decrease with time, while below the balance point, the concentration will increase.

While barrel storage under dry conditions has the undesirable effect (especially if near 14%) of raising the ethanol concentration, it provides the benefit of concentrating aromas and flavors somewhat. Barrel storage under high humidity conditions, however, allows fruit to be picked later with riper flavors; the resulting, higher alcohol wines will then lose some of the alcohol during storage.