Engineer G. I. FUKS

ON THE QUESTION

OF THE PRODUCTION OF DRY ICE

AND ITS APPLICATION FOR THE NEEDS

OF TRANSPORT

Tomsk

1929

Separate offprint from Izv. STI, vol. 51, issue IV.

Okrlit 846. Tomsk, printing house “Kr. Zn.” Order No. 2474-30. Print run 550 copies.

Eng. G. I. Fuks.

On the Production of Dry Ice and Its Application for Transport Needs

Theoretical grounds for the question of the production of dry ice, its cost, and the cost of refrigeration during transport¹.

This article was written at the suggestion and on the idea of Prof. of the STI S. V. Lebedev, as part of a general work on the production and application of dry ice for the railway transport of export Siberian butter.

The development of the export of perishable products and, in particular, of the export of butter from the Siberian butter-making regions rests, as is known, on the question of creating a sufficiently advanced refrigeration chain. One link of this chain in particular is imperfect—refrigerated transport. A product that is very valuable and of good quality at the place of manufacture loses its qualities in transit and, as a result, is valued below the commodity prices of other countries.

Proceeding from the obvious proposition that the development of exports of perishable products is, for the USSR, the most urgent task of the time, one must pay special attention to all proposals in the field of improving the conditions of refrigerated transport. Nor should one be embarrassed by the fact that the proposed arrangement will be somewhat more expensive than that adopted up to now. In relation to the product being transported itself, its transport is only a small overhead expense. Even a considerable, severalfold increase in this expense, if only this expense is sufficiently expedient, may be justified. It is necessary to connect the whole question with the economic side, i.e., to be convinced that the additional expenditure incurred will produce a definite commercial effect—a higher price for the transported product at the place of sale, more than covering the additional costs.

The history of the question of systems of refrigerated transport is quite extensive. Attempts have repeatedly been made to replace the ordinary ice refrigerator car with a more advanced device. Refrigeration was proposed according to an ice-salt system with artificial circulation of brine and air (the cars of Benneter, Sokolov and Ptashinsky, etc.); finally, Linde’s mechanically refrigerated train was tested on our railways.

The matter did not proceed beyond trial shipments, which sometimes yielded very good results in the sense of preserving the quality of the product being transported. The reasons for this are usually considered to be:

1. The high cost of transport in comparison with an ordinary ice refrigerator car.

1. The complexity of setting up all these refrigeration systems, i.e., fairly expensive and complicated equipment which, in essence, can be used only periodically and for a comparatively short time during the year.

2. The need for careful and qualified servicing of the installation, which is difficult and expensive under ordinary transport conditions.

The first reason, under the conditions indicated above, is not as important as it may seem at first glance. It must be assumed that the decisive role was played by reasons 2 and 3.

In this respect, attention is deserved by a new method of refrigeration now being used in America. This method of refrigeration consists in the use of “dry” ice, or solid carbon dioxide, instead of ordinary ice for refrigeration purposes.¹)

The idea of using solid carbon dioxide for refrigeration purposes cannot be considered entirely new. For the first time, as far as the author knows, it was proposed to use carbon-dioxide snow for cooling fish in fishing boats in Russia in 1910–12.

On the other hand, the proposal to make use of solid carbon dioxide for purposes of ordinary refrigeration cannot but seem strange at first glance. The practice of storing and transporting perishable products requires moderate refrigeration, from 0 to \(-5^\circ\)C, at most down to \(-20^\circ\)C. Solid carbon dioxide in the form of ice or snow evaporates at atmospheric pressure at about \(-79^\circ\)C. Since the cost of each calorie of cold increases greatly as the temperature is lowered, it is clear that using solid carbon dioxide to obtain moderate refrigeration will be uneconomical, at least where these moderate cold temperatures can be obtained directly, for example by means of a refrigerating machine.

Nevertheless, American practice has revealed a number of advantages in the use of “dry” ice for refrigeration purposes.

1. Dry ice does not melt, but passes directly into vapor (gas), without any residue. The temperature of evaporation in this case is \(-78.9^\circ\)C, so that part of the cold can still be obtained when the resulting gaseous carbon dioxide is warmed. When dry ice is used, no liquid is formed, which justifies its name, i.e., there is no excess moisture in the refrigerated space.

2. Dry ice is approximately twice as heavy as ordinary ice.

3. The gaseous carbon dioxide formed from dry ice counteracts putrefactive processes.

4. Dry ice has, per unit of weight, approximately twice the refrigerating capacity of ordinary ice. Per unit of volume occupied, its action is, therefore, four times greater.

To this it remains to add that the low temperature of the cold obtained makes it possible to count quite reliably on obtaining sufficiently low temperatures in the refrigerated space (car).

¹) LITERATURE: Bulletin mensuel de renseignements frigorifiques, 1927, 1, 2.
Martin. “The Field of Dry Ice in Modern Refrigeration.” Refrigerating Engineering, 1928, 2.
Taylor. “Refrigeration of Fish. Solid Carbon Dioxide Refrigerant for Transportation.” Bulletin 1016, Department of Commerce, Washington.
Rulev. “Carbon Dioxide, or Dry Ice, in the Fish Industry.” Perishable Products and Refrigeration, 1928, 7.
— “Production of Dry Ice from Carbon Dioxide.” Ibid., 1928, 8.
Krause. “Kohlensäure als Kältemittel.” Z. für d. gesamte Kälte-Industrie, 1928, 10.

The possibility of using dry ice for cooling purposes may be considered proven, especially since it has been proven by shipments of frozen fish in America. Shipments were carried out in isothermal cars from Halifax and Montreal over the course of 6 days; in addition, a trial shipment was made of fish in the form of frozen fish fillets, packed together with dry ice in the hold of an ordinary unrefrigerated vessel from New York to Kingston (island of Jamaica) over the course of 7 days. In both cases the transported cargo arrived in good condition.

As for the question of the economic advantage of using dry ice under the conditions of our transport, it comes down first of all to the problem of the cost of dry ice.

In view of the novelty of this matter, it is not possible to obtain firm data on the price of dry ice. One can only outline approximate figures and compare them with the few data that are available in the literature.

The technique for obtaining solid carbon dioxide, or dry ice, is fairly simple. Having liquid carbon dioxide in a cylinder or at the outlet from a condenser at a pressure of 50–70 absolute atmospheres and a temperature of 10–25° C, it is passed through a narrow throttling orifice, calculated so that the pressure of the carbon dioxide drops to normal atmospheric pressure. Part of the carbon dioxide evaporates in this process, while another part passes into the solid state in the form of snow. The Dry Ice Corporation uses as an evaporator a steel cylinder about 900 mm in diameter and about 1500 mm high, carefully insulated on the outside with cork. The evaporator proper is a second steel cylinder, inserted concentrically into the first with a gap. This inner cylinder is covered at the top with cloth and a sieve, through which the gaseous carbon dioxide is drawn off.

When the inner cylinder is filled with carbon-dioxide snow, the latter is raked out through a special opening and pressed by a hydraulic press under a pressure of 55–100 atm. In this way dry ice of density 1.1 is obtained.

The main question in calculating the cost of dry ice is the question of the yield of dry ice in this process. Engineer V. N. Rulev indicates that in ordinary laboratory experiments only 10–15% of the liquid carbon dioxide is obtained in the form of snow, while in industrial enterprises 1/3 and more of snow is obtained. Since the yield of solid carbon dioxide is the principal basis for calculating its cost, we shall in what follows determine a more exact figure for its possible yield on the basis of the corresponding theoretical data. For this we shall use the data of two works: R. Plank: Herstellung und Industrielle Verwendung fester Kohlensäure¹), and R. Plank and J. Kuprianoff: Die thermische Eigenschaften der Kohlensäure in gasförmigen, flüssigen und festen Zustand²).

The calculation has been made for a specific installation with a capacity of about 500 poods (~8000 kg) per day. As the source of carbon dioxide we shall assume carbon dioxide obtained in the process of fermentation, i.e. a gaseous waste product of production, usually not utilized. The process of obtaining solid carbon dioxide is based—

¹) VDI, 1929, 7.
²) Z. f. d. ges. Kälte-Industrie, 1929, 3.

...the process of three-stage compression with likewise three-stage evaporation; moreover, in the intermediate evaporators (subcoolers) we assume heat exchange between the compressed vapor and the liquid flowing in through the throttle valve (see the installation diagram, drawing No. 2). Thus, we have an approximation to the most advantageous regenerative process of the installation¹.

Diagram
of an installation for obtaining solid CO₂.

[Drawing No. 2]

Visible labels in the diagram:

• A. Compressor, 1st stage; 1 atm.—6 atm.
• B. Compressor, 2nd stage; 6 atm.—20 atm.
• C. Compressor, 3rd stage; 20 atm.—65 atm.
• E. Condenser.
• F. Liquid CO₂ subcooler.
• G., H. First and second intermediate coolers (liquid separators).
• R, R′, R″. Regulating valves.
• H₁, H₂. Usual coolers of compressed CO₂ between stages (water).
• Q. Generator of solid CO₂ (“snow”).

Temperature and pressure markings visible in the diagram include: \(+65^\circ\), \(+25^\circ\), \(+15^\circ\), \(-20^\circ\), \(-50^\circ\), \(-80^\circ\); 20 atm., 6 atm., 1 atm.

Legend:

• F — Subcooler of liquid CO₂.
• G, H — First and second intermediate coolers (liquid separators).
• R, R′, R″ — Regulating valves.
• H₁, H₂ — Usual coolers of compressed CO₂ between stages (water).
• solid line — Pipelines for vapor/gas.
• dashed line — Water lines.

Practical deviations from it shall be taken into account by the corresponding coefficients, as is customary in the practice of calculating refrigerating installations.

Using the entropy diagram for carbon dioxide, we plot the theoretical process of operation of the compressor. The intermediate pressures are taken as 6 atm. and 20 atm. abs., i.e., −50° C and −20° C (Fig. No. 1).

Entropy diagram of CO₂

[Figure No. 1. Entropy diagram of CO₂]

Theoretical process.¹)

When the liquid is throttled through the first regulating valve \(R'\), according to the diagram the following is obtained:

0.41 kg of vapor and 0.59 kg of snow.

The vapor is obtained at a pressure of 1 atm. abs. and \(t = -80^\circ\text{C}\).

For continuity of the process, it is necessary to add to it 0.59 kg of vapor, whose temperature we shall take as equal to the temperature of the first intermediate cooler, \(-50^\circ\text{C}\).

The mean temperature of the vapor drawn in by the first stage will therefore be:

\[ t = \frac{0.41(-80) + 0.59(-50)}{1} \sim -60^\circ \text{C} \]

¹) Note: All characteristic points of the process have double numbering, the second digit indicating the number of the stage.

Meanings of the first digit:

1—beginning of compression
2—end of compression

3—end of condensation
4—beginning of evaporation

For example: \(i_{43}\)—heat content of the mixture at the beginning of evaporation in the 3rd stage (2nd intermediate cooler).

When compressing the vapor to a pressure of 6 atm. in the 1st stage, the theoretical work expenditure required is:

\[ A L_{ad} = i_{21} - i_{11} = 182 - 157 = 25 \ \text{Cal} \]

For the complete conversion into liquid of the vapor compressed in the 1st stage, from each kg it is necessary to remove:

\[ Q_{21} = i_{21} - i_{31} = 182 - 72 = 110 \ \text{Cal} \]

Since at the end of compression the vapor is at a high temperature \((+68^\circ \mathrm{C})\), part of the heat can be removed from it by cooling water. Assuming cooling to \(25^\circ \mathrm{C}\) \((i = 173)\), we find that the cooling water will remove:

\[ q_2' = 182 - 173 = 9 \ \text{Cal/kg}. \]

Remainder:

\[ Q_{21} - q_2' = 110 - 9 = 101 \ \text{Cal/kg}. \]

must be removed by evaporation of the liquid in the 1st intermediate cooler. Thus, into the 1st intermediate cooler there enters a mixture of liquid and vapor with heat content \(i_{32} = i_{41} = 89 \ \text{Cal}\), and dry saturated vapor with heat content \(i_{12} = 155 \ \text{Cal}\) is drawn off; therefore, by one kg one can remove heat:

\[ Q_{12} = i_{12} - i_{42} = 155 - 89 = 66 \ \text{Cal/kg}. \]

It is also necessary to take into account that in the first intermediate cooler 0.59 kg of vapor must be cooled from \(-20^\circ \mathrm{C}\) to \(-50^\circ \mathrm{C}\), for which it is necessary to remove

\[ q_1 = 0.59 \cdot 0.2 \cdot 30 = \sim 3.5 \ \text{Cal} \quad (\text{mean } \sim 0.2) \]

Consequently, in all, in the 1st intermediate cooler it is necessary to remove

\[ (Q_{21} - q_2') + q_1 = 101 + 3.5 = 104.5 \ \text{Cal/kg}. \]

Per one kg of carbon dioxide in the 1st stage, the following must come into the second:

\[ g_2 = \frac{(Q_{21} - q_2') + q_1}{Q_{12}} = \frac{104.5}{66} = 1.58 \ \text{kg}. \]

Further analogous calculations, as well as the computed values for the first stage, are collected in Table No. 1.

Table No. 1.

The following additions must be made to this table:

The amount of heat that must be removed in the second intermediate cooler in order to cool the additional gas is calculated, assuming cooling in it from \(+25\) to \(-20^\circ\mathrm{C}\):

\[ q' = C_p G \cdot \Delta t = 0.2\; 0.59 [25 - (-20)] \sim 5.3\ \mathrm{Cal/kg}. \]

But since for each kg of working substance of the 1st stage, \(1.58\) kg works in the second, then for each kg of working substance of the 2nd stage this will be:

\[ \frac{5.3}{1.58} \simeq 3.4\ \mathrm{Cal/kg}, \]

which is what is entered in the table.

The amount of working substance that must pass through the second stage per 1 kg of the first stage has been calculated above—\(1.58\) kg. Accordingly, the amount of working substance in the 3rd stage, falling per 1 kg of the 2nd stage, is:

\[ g_3 = \frac{83.4}{49} \simeq 1.72\ \mathrm{kg}. \]

Actual process.

First stage.

The actual process will differ from the theoretical one just considered by a number of losses of a thermal and mechanical character.

As a result of heat penetration, in order to obtain the specified effect it will be necessary to pass a greater quantity of working substance.

We shall take, according to Koeniger:

\[ \rho = 1.2 \quad \text{for the first stage,} \]

i.e., we shall assume that, to compensate for losses from heat penetration in the 1st stage, it is necessary to pass \(20\%\) more working substance than in the theoretical installation.

For 1 kg of carbon-dioxide snow, in the 1st stage it is necessary to pass:

\[ g' = \frac{\rho}{0.59} = 2.035\ \mathrm{kg}. \]

We shall estimate the compressor losses by the coefficient:

\[ \eta_d = 0.8 \]

The cooling of the added vapor in the 1st intermediate cooler, for the possibility of heat transfer, we shall take from \(-10^\circ\) to \(-40^\circ\), i.e. by \(30^\circ\mathrm{C}\), for which it is necessary to expend \(\sim 3.5\) Cal.

The initial compression temperature is

\[ t = \frac{0.41(-80) + 0.59(-40)}{1} = -57^\circ. \]

Plotting this point on the \(T S\) diagram and drawing the adiabat to a pressure of 6 atm., we read off, by the adiabat, the compression work:

\[ A L_{ad} = 183.5 - 158 = 25.5\ \mathrm{Cal/kg}. \]

Actual compression work per 1 kg:

\[ A L_d = \frac{A L_{ad}}{\eta_d} = \frac{25.5}{0.8} = 31.9\ \text{Cal/kg}. \]

Actual compression work of the 1st stage per 1 kg of carbon-dioxide snow:

\[ A L_1' = A L_d' = 31.9 \cdot 2.035 = 65.0\ \text{Cal/kg}. \]

Heat content at the end of compression in the 1st stage:

\[ i_2 = i_2 + A L_d = 158 + 31.9 = 189.9\ \text{Cal}. \]

By cooling to \(25^\circ C\), the following amount of heat can be removed:

\[ 189.9 - 173 = 16.9\ \text{Cal/kg} \]

by the cooling water. Per 1 kg of dry ice, therefore, the water must remove:

\[ 16.9g' = 34.4\ \text{Cal/kg}. \]

For further complete conversion into liquid, it is necessary, in addition, to remove:

\[ 173 - 72 = 101\ \text{Cal/kg}. \]

That is, per 1 kg of agent in the 1st stage, \(1.58\) kg must theoretically be passed into the second; in practice, however:

\[ 1.58\rho = 1.58 \cdot 1.2 = 1.9\ \text{kg}. \]

And per 1 kg of solid carbon dioxide:

\[ \frac{1.90}{0.59} = 3.21\ \text{kg/kg}. \]

Estimating the thermal losses in the intermediate cooler by a coefficient of \(1.2\), it is necessary in the second stage to pass, per 1 kg of solid carbon dioxide:

\[ g = 3.21 \cdot 1.2 = 3.86\ \text{kg/kg}. \]

The corresponding calculations for the second and third stages are given in Table No. 2. The same table also includes the data calculated above for the first stage.

Table No. 2.

The total amount of work expended on compression, per 1 kg of snow obtained:

\[ AL = 65.0 + 72.3 + 109.5 = 247 \ \text{Cal/kg}. \]

The total amount of heat transferred by the cooling water per 1 kg of snow is \(\sim 530.0\) Cal/kg.

The data necessary for calculating the dimensions of the compressor are summarized in Table No. 3.

Table No. 3

Using the data of Tables Nos. 2 and 3, we calculate the dimensions and power consumption of an installation with a capacity of \(G = 500\) poods per day, \(\sim 8000\) kg/day of dry ice. Assuming 20-hour operation of the installation, we have the required indicated power:

\[ N_i = \frac{AL\,G/20}{632 \cdot 3} = \frac{400 \cdot 247}{632 \cdot 3} \sim 156 \ \text{hp}. \]

The mechanical efficiency of the compressor is taken as \(\eta_m = 0.85\). Then the required effective power is:

\[ N_e = \frac{N_i}{\eta_m} = \frac{156}{0.86} \sim 184 \ \text{hp}. \]

Allowing for losses in transmission, a motor must be installed of

\[ N = 200 \ \text{hp}. \]

The total hourly consumption of water, assuming its heating by \(\sim 14^\circ\text{C}\),

\[ G_v = \frac{530 \cdot 0 \, G/20}{14} \sim 15 \ \text{m}^3/\text{hour}. \]

To calculate the dimensions of the compressor, we compute the required hourly volume swept by the compressor piston according to the formula

\[ V_r = V G/20 \]

where \(V\) is taken from Table 2.

As a result we have:

Hourly volume swept by the piston of the 1st stage: \(524 \cdot 10^6\) cm.\(^3\)
” ” ” ” 2 ” \(116 \cdot 10^6\)
” ” ” ” 3 ” \(75 \cdot 10^6\)

In order to obtain an idea of the actual dimensions of the compressor, we adopt the following scheme of its construction (Fig. 3).

Assuming the number of revolutions per minute \(n = 300\) rev/min and the compressor stroke \(S = D_3\), we obtain the following compressor dimensions:

\[ D_1 = 495\ \text{mm.}\quad D_2 = 410\ \text{mm.}\quad D = 175\ \text{mm.} = s. \]

A compressor of the indicated dimensions, according to an approximate calculation, should cost approximately 15,000 rubles.

Calculation of the Cost Price of 1 kg of Dry Ice

Starting from the established concrete dimensions, we draw up an approximate estimate of the initial costs for the installation.

The initial costs of the installation, according to the corresponding approximate estimate, amount to:

Cost of the building . . . . . . . . . . . . . . . . 18000 r.

Cost of machinery equipment (compressors, electric motors, presses, etc.) . . . . . 144500 r.

Approximate estimate of annual expenses.

Operating expenses:

The operation of the installation is assumed to be seasonal, for 100 days a year, 20 hours per day. Drive by an electric motor \(N = 200\) h.p. \(\sim 150\) kW. The efficiency coefficient of the motor is taken as 0.95, the cost of current is 8 kopecks per kilowatt-hour. The annual expense for electric power is:

\[ \frac{0.08 \cdot 150 \cdot 20 \cdot 100}{0.95} = 25250\ \text{r.} \]

The annual expense for water, with its consumption of 15.0 m.\(^3\)/hour and a price of 25 kopecks per cubic meter:

\[ 0.25 \cdot 15.0 \cdot 20 \cdot 100 = 7500\ \text{r.} \]

Lubricating and wiping material . . . . . . . . . . . . . 500 r.

For servicing the installation the following personnel are assumed:

1 shop superintendent
3 production foremen

4 machinists (3 shifts and 1 reserve)
4 press operators
4 press-operator assistants
7 unskilled laborers.

Valuing the labor force according to the wage schedules, we obtain the total annual expenditure for the service personnel . . . 22,800 rubles.

The total amount of annual operating expenses, including also 15% for unforeseen expenses, will be . . . 64,500 rubles.

The total gross expenses for the installation, counting depreciation of the building at 5% and depreciation of the machinery at 10%, will be . . . 80,000 rubles.

The production of dry ice is 400 kg/hour; counting 20 hours for 100 days per year:

\[ 400 \cdot 20 \cdot 100 = 800{,}000\ \text{kg/year} \]

Counting 25% for evaporation before reaching the consumer, we have useful production:

\[ 0.75 \cdot 800000 = 600000\ \text{kg/year}. \]

Cost of 1 kg of dry ice:

\[ 80000 : 600000 = 0.1345\ \text{rubles} \sim 13.5\ \text{kopecks/kg}. \]

Taking into account the cost of the principal product (gaseous carbon dioxide) at 1.6 kopecks/kg (the cost of purification, etc.), we obtain the full cost of 1 kg of dry ice as 15 kopecks per kilogram.

This cost figure is considerably (~1.5 times) lower than those cited in the literature from American practice. Since, nevertheless, the reality of the figure presented, as an approximate one, is not subject to doubt, it remains to propose two sources of the discrepancy:

1. Imperfection of the working process in the installations that have been built. The installations that have been built, so far as is known, operate with compression in a single stage, which must considerably reduce the yield of product for approximately the same expenditure of work.
2. High commercial profit.

Application of Dry Ice in Transport

Having calculated the cost of 1 kg of dry ice, let us try, as an approximation, to take into account the question of its operational use as a source of refrigeration in transporting export butter from Siberia.

At atmospheric pressure dry ice does not melt, but passes directly into vapor at −78.9° C, with the latent heat of sublimation being 136.89 cal/kg. The resulting carbon-dioxide vapors will have the same temperature, −78.9° C. If measures are taken to avoid the loss of cold vapors, i.e., to use the resulting vapor for cooling purposes, then a certain additional quantity of refrigeration can be obtained. Since for the needs of refrigerated transport we usually require a temperature only slightly below (by 2–3° C) zero, it may be calculated that the vapor will be discharged at a temperature of approximately −10° C.

To calculate the heat that will be required to warm the CO₂ vapors from −78.9° to −10°, it is quite sufficient to use an approximate equation for the heat capacity, treating CO₂ as an ideal gas.

We take, according to Planck:

\[ C_p = 0.1965 + 0.00023\,t. \]

Then, upon being heated from −78.9° to −10°C, 1 kg of CO₂ will require:

\[ q=\int_{-78.9}^{-10} C_p\,dt = \left[ 0.1965+0.00023\,\frac{-10-78.9}{2} \right] [-10-(-78.9)] = 12.85\ \mathrm{cal}. \]

Consequently, in all, 1 kg of dry ice will yield cold:

\[ 136.89+12.85 = 149.74 \sim 150\ \mathrm{cal./kg}. \]

Let us note that this amount of cold is approximately twice as great as that which 1 kg of ordinary ice can provide; moreover, the cold itself is obtained at a considerably lower temperature.

Further, in a wagon cooled by dry carbon-dioxide ice, a very harmful factor for the storage of any perishable product is completely eliminated, namely the excessive dampness or humidity of the air. The following simple calculation convinces us of this:

Each kg of dry ice, on evaporating, increases considerably (approximately 570-fold) in volume. At atmospheric pressure, 1 kg of carbon-dioxide vapor occupies a volume of 365.12 dm at \(t=-78.9^\circ\)C. Upon further heating to −10°C, the volume occupied by 1 kg must increase in proportion to the absolute temperatures, i.e.:

\[ \text{by }\frac{273-10}{273-78.9}\sim 1.35\text{ times}. \]

Thus, the volume of one kg of the carbon dioxide evolved will be approximately 495 dm³.

If it is assumed that approximately 1500 kg of dry ice will be loaded into the wagon, then, upon evaporation and heating, its vapor must occupy a volume of

\[ 1500 \cdot 0.495 \sim 740\ \mathrm{m.^3}. \]

The total volume of an ordinary insulated wagon is approximately 38.5 m³. Thus, the volume of the vapor obtained exceeds the volume of the wagon by approximately 19 times.

Taking into account the fact that during the run the wagon must be closed as tightly as possible and that, consequently, the gas will escape from it only through the unavoidable leaks in the lining and insulation, one may expect that most of the air present in the wagon during the loading of the dry ice will be displaced. In this way, the admixture of water to the air will also be displaced. In the wagon, soon after loading, a fairly high content of carbon dioxide should be established. It is, of course, necessary to carry out a detailed investigation of the influence of an atmosphere rich in carbon dioxide on the preservation of one or another product. But a priori it is difficult to expect that the presence of such a chemically indifferent gas as CO₂ could be harmful, for example, to butter, meat, etc.¹

¹ The question of fruits, etc., of course, stands apart.

The basic question of the cost of cooling en route when using dry carbon-dioxide ice we shall try to illuminate by the following trial calculation for the run of an ordinary loaded oil car, Novosibirsk—Leningrad. It is necessary only to stipulate in advance that, for the purposes of a preliminary, approximate calculation, considerable simplifications will be introduced into the calculation, for the most part in a direction unfavorable to the new method of cooling, in order to have the most reliable quantities for comparison.

The average commercial speed of movement of an isothermal car is taken, according to data of the Tomsk Railway for the 1927/28 operating year, as \(W = 20\) km/hour.¹ Taking the distance Novosibirsk—Leningrad as \(S = 3620\) km, we have the time the car remains en route, in hours:

\[ \mathrm{n} = \frac{S}{W} = \frac{3620}{20} = 131 \text{ hours, i.e. } \sim 7.5 \text{ days} \]

For the comparative calculation we shall take the necessary running time of the car Novosibirsk–Leningrad at the speed of a goods-passenger train—6 days. Counting, as a minimum, 4 packings of ordinary ice en route for replenishment, we have a total running time of 7 days. To this must be added:

12 hours from the beginning of the packing of the ice to delivery to the refrigerator with the loading of oil in Novosibirsk.

12 hours from the completion of the loading of the oil to departure from Novosibirsk.

24 hours from the moment the car arrives in Leningrad to unloading—altogether 2 days.

Thus, the total time of stay of the isothermal oil car under cooling is determined as 9 days, or \(n_0 = 216\) hours (when cooled with ordinary ice).

When cooled with dry ice this time may be reduced by:

1) 12 hours in N.-Sibirsk, i.e. preliminary cooling of the car is not required.

2) 24 hours en route, since the need for replenishing the ice disappears.

In all, by 36 hours, which gives a time of stay under cooling \(n_c = 180\) hours.

The coefficient of thermal conductivity of the car walls is taken in the further calculation as \(k = 0.45\). This figure is taken as a certain average for the whole car, after calculating the average coefficient of thermal conductivity of a number of isothermal cars described in the literature, and also according to data of engineer P. I. Denisov.² This coefficient is somewhat less than that which was obtained with prewar insulation of an oil car with cork alone, although it was reached in other cars. With modern types of car insulation using shevelin, morozin, felt, etc., the coefficient of thermal conductivity usually does not exceed the indicated value.

There is no doubt³ that the coefficient of thermal conductivity of the car walls should be chosen from economic relations, as the most advantageous. In this case one must expect that cars with cool—

cooling with dry ice must be provided with more perfect insulation, with a lower coefficient of thermal conductivity, than cars cooled with ordinary ice. By adopting the same coefficient of thermal conductivity for both types of cars, we thus overstate the consumption of dry ice, i.e., we calculate with a certain margin.

The outside-air temperature is taken as \(22^\circ\mathrm{C}\), and the temperature inside the car when cooled with ordinary ice as \(+2^\circ\mathrm{C}\), when cooled with dry ice as \(-2^\circ\mathrm{C}\). Thus the calculated temperature difference when cooled with ordinary ice is:

\[ \Delta t_o = 22 - 2 = 20^\circ\mathrm{C}. \]

When cooled with dry ice:

\[ \Delta t_c = 22 - (-2) = 24^\circ\mathrm{C}. \]

For the butter exporter, the difference from \(+2\) to \(-2^\circ\mathrm{C}\) in the storage and transport of butter is so obvious that it is superfluous to dwell on it. The thermal calculation, of course, does not take into account the “quality” of the cooling temperature, but there is no doubt that, with so low a temperature of the main cooling agent as dry ice \((-78.9^\circ\mathrm{C})\), attaining a temperature of \(-2^\circ\mathrm{C}\) inside the car is considerably easier than attaining \(+2^\circ\mathrm{C}\) when cooling with ordinary ice, whose melting temperature is about \(0^\circ\mathrm{C}\).

We shall carry out the calculation for an ordinary “butter” car,¹ the basic data on which are given in Table No. 4.

Table No. 4.

Cooling devices: number—8, at the end walls; dimensions

\[ 0.410 \times 0.410 \times 1.500\ \mathrm{m}^3. \]

The required amount of refrigeration for cooling the car en route, net, when cooled with ordinary ice:

\[ Q_o^n = \kappa F \Delta t_o n_o = 0.45 . 85.8 . 20 . 216 = 167000\ \mathrm{Cal} \]

The same when cooled with dry ice:

\[ Q_c^n = \kappa F \Delta t_c n_c = 0.45 . 85.8 . 24 . 180 = 167000\ \mathrm{Cal}. \]

In determining the quantity of ice, it is necessary, in addition to this net expenditure of refrigeration, also to take into account a number of various losses of refrigeration, sometimes very considerable. In order to obtain an idea of the magnitude of these losses, the results of two experimental shipments were used: the experimental shipment Astrakhan–Petrograd¹ and the experimental trip Kurgan–Leningrad in the summer of 1924.² In this connection, by planimetering the ter-

grams, the average temperature difference during the run was determined, the necessary net quantity of ice was calculated, and it was compared with the quantity actually consumed. The results of the calculations are summarized in Table No. 5.

Table No. 5.

The utilization coefficient was calculated as the ratio:

\[ \eta = \frac{\text{Required quantity of ice}} {\text{Consumed quantity of ice}} \cdot 100\% \]

On the basis of these data, the utilization coefficient for cooling with ordinary ice is taken to be 50%.

The necessary quantity of ordinary ice $G_0$, taking, as usual, for cooling with salt a latent heat of fusion of 75 Cal./kg:

\[ G_0 = \frac{Q_0^{\mathrm{n}}}{\eta \cdot 75} = \frac{167000}{0.50 \cdot 75} \sim 45000\ \text{kg}. \]

In addition to this, there is the consumption of salt, taking on average 10–12% by weight of the ice charge:

\[ g = 0.11\,G_0 = 0.11 . 4500 = 490\ \text{kg}. \]

The cost of procuring ice, according to the data of Eng. P. I. Denisov¹ for the 1923/24 operating year, averaged 20 r. 30 k. per 1 cubic sazhen, with considerable fluctuations from 32 r. 90 k. to 6 r. 90 k.

In view of the large fluctuation of these figures and their considerable instability from year to year, for the calculations the average cost of ice according to the data of the Tomsk Railway for the 1927/28 operating year has been adopted. These data are as follows.

Cost of procuring 1 m³ — 1 ruble.

Cost of loading into a car, 1 t — 2 rubles.

“ salt, 1 t — 26 rubles.

According to these data, the cost of 1 kg of ice loaded into the car is calculated as follows. The percentage of ice melting during procurement, according to the data of Eng. P. I. Denisov, is 25–50%. For Siberian conditions pri-

a figure close to the lower limit of 30%. Taking the specific gravity of the ice in the blank as 0.7, and the cost of loading as 2 rubles, we have the cost of 1 kg of ice in the car:

\[ a_0=\frac{\frac{100}{0.7\cdot(1-0.3)}+200}{1000}=0.404\ \text{kopecks/kg}. \]

The cost of all the ice for cooling, including loading:

\[ a_0G_0=0.404\cdot4500=18\ \text{rub.}\ 20\ \text{kop.} \]

The cost of salt:

\[ \frac{2600}{1000}g_0=2.6\cdot490=12\ \text{rub.}\ 70\ \text{kop.} \]

The total cost of cooling en route:

\[ 18\ \text{rub.}\ 20\ \text{kop.}+12\ \text{rub.}\ 70\ \text{kop.}=30\ \text{rub.}\ 90\ \text{kop.} \]

The type of car we have adopted holds \(\sim150\) barrels, or 8 t of butter. The en-route expenditure for cooling one ton of butter is

\[ K_0=\frac{30\ \text{rub.}\ 90\ \text{kop.}}{8}=3\ \text{rub.}\ 86\ \text{kop.} \]

When calculated per 1 t of butter per day, this gives a cooling cost of 43.0 kop./t. This figure considerably (by \(\sim 5\) times) exceeds the figure given in Hütte¹ of 8.1 kop./t; moreover, the latter figure includes all expenses for refrigerated transport. It is obvious that the latter figure was calculated for other (probably doubled) ratios, since the cost of the ice alone in the blank already gives, when calculated per 1 t of butter, a larger figure.

When cooling with dry ice, we find its required quantity from the condition that 1 kg of dry ice, evaporating and warming up, absorbs 150 Cal./kg. In this case, a significantly better utilization of the cold in the car should be expected, since a number of factors that worsen the utilization of the cold are eliminated. In fact, the required amount of dry ice for cooling the car is small and can be loaded into the car at once, at the departure station, without the need to replenish it later. Then, in the construction of the car, places where leaks are most likely (loading-hatch covers, siphon openings, etc.) are eliminated. In view of the considerable refrigerating capacity of dry ice and its placement in the car above the product being cooled, the reserve of dry ice by the end of the run can be made minimal, which, as is known, is not permissible with ordinary cooling. Experimental shipments in America with dry ice, which were mentioned above, confirm this result. The amount of dry ice loaded in them was equal to, or only slightly exceeded, the theoretically necessary amount. Nevertheless, for caution we introduce the coefficient of utilization of the car when cooling with dry ice:

\[ \eta_c=0.75. \]

The required quantity of dry ice will then be determined as

\[ G_c=\frac{Q_c^{\mathrm{n}}}{\eta_c 150} =\frac{167000}{0.75\cdot150}\sim1500\ \text{kg}. \]

The price of dry ice was determined above as 15 kopecks/kg, or 150 rubles per ton. The cost of loading dry ice should not exceed the cost of loading ordinary ice. Let us take the cost of loading 1 ton of dry ice as 3 rubles; then the total cost of cooling the car en route will be:

\[ 1.5\cdot(150+3)\sim230\ \text{rubles}. \]

In absolute magnitude this figure is approximately 6.5 times greater than the cost of cooling by the ordinary method—ice with salt. Thus, there can be no question of applying this method of cooling where the ordinary ice-and-salt method is sufficiently satisfactory. But in those cases where this is not so, the chief role will be played not by the absolute cost of cooling, but by the difference in expenses per unit of weight and the difference in selling price.

The cost of cooling 1 ton of butter with ordinary cooling was indicated above. In calculating the cost of cooling per unit of weight with dry ice, it is necessary to take into account the increase in the capacity of the car when cooling with dry ice. This increase in capacity is due to two causes.

1. The need for cooling pockets at the end walls of the car, which take up a rather considerable area that can be used for loading goods, is eliminated. Calculation according to the data given above gives a gain in area of approximately 16%.

There is the possibility of considerably better utilization of the cubic capacity of the car, since there is no need to leave so great a distance between the tops of the barrels and the ceiling of the car as is done with ordinary cooling for air circulation.

A certain volume of the car, it is true a very small one, will be used for placing boxes of dry ice. But they are placed above, just under the ceiling, where loading of the transported cargo is in any case impossible.

Taking, as a minimum, the gain in loading of the car proportional to the gain in area freed from the cooling apparatus—approximately 15%—we obtain, for loading with dry ice, the weight of transported butter:

\[ 1.15\cdot8.0=9.2\ \text{t}. \]

The expense en route Novosibirsk—Leningrad for cooling one ton of butter is:

\[ K_c=\frac{230}{9.2}=25.0\ \text{rubles/t}. \]

In comparison with cooling by ordinary ice with salt, cooling with dry ice will be more expensive by

\[ 25.0-3\ \text{rub.}\ 86\ \text{kop.}=21\ \text{rub.}\ 14\ \text{kop./t},\quad \text{or}\sim35.2\ \text{kop./pood}. \]

This calculation requires more or less significant corrections, which give some reduction in the computed difference in cost. Without including them in the figures, let us indicate the principal ones:

1. Reduction in the turnaround time of the car—by approximately 1 day.

1. Better utilization of the car’s lifting capacity.

2. Simplification, and perhaps even reduction to a minimum, of en-route servicing of the car (absence of repeated re-icing).

It can only be noted that, even without taking into account the indicated corrections, the use of dry ice for cooling cars may be commercially profitable in the transport of export butter. One must suppose that the possibility of steadily maintaining the necessary cooling temperature for butter en route sufficiently guarantees the preservation of the butter’s quality; transport experience shows that butter loses most of all in quality during prolonged transport in a poorly cooled car.

According to data of Eng. P. I. Denisov,¹ relating to 1924, the differences in prices on the London market between Siberian and Australian butter are expressed as 2 rubles/pood—120 rubles/ton; between Siberian and Danish butter, up to 5 rubles/pood, i.e., 300 rubles/ton. If one calculates that, owing to improved transport conditions, the quality of Siberian butter on the London market would at least approach the quality of Australian butter, then one may count on an increase in the selling price of 100–80 rubles/ton. Deducting 20 rubles of additional expenses per ton, caused by the use of more expensive dry ice for cooling, we have a real gain per ton of 60–80 rubles.

According to the statistics of 1913, the export of butter from Siberia was 83 thousand tons, of which 78,000 went abroad. Under these conditions the real benefit from the use of dry ice should be expressed as the sum:

\[ (60\text{–}80)\ 78000 = 4.68 \text{ to } 6.24 \text{ million rubles/year}. \]

Summarizing all that has been set forth, we arrive at the following conclusions:

1. Dry carbon-dioxide ice, by virtue of its properties: a) high specific gravity, b) low evaporation temperature, and c) direct conversion into vapor (under normal conditions) without an intermediate liquid phase, is a very convenient and compact source of cold. Cold can be obtained at temperatures considerably lower than in the melting of ordinary ice, even with an admixture of salt, and, in any case, at a temperature fully satisfying the strictest requirements in this respect.

2. The obtaining of the principal product—gaseous carbon dioxide—is not associated with any special difficulties. Gaseous carbon dioxide may be obtained from the products of fuel combustion under steam boilers, as the Dry Ice Co. does in America, or as a by-product of fermentation. In both cases the matter is reduced to a simple and inexpensive washing of the gas in order to extract from it, if possible, pure CO₂.

3. The technical execution of all the necessary apparatus for obtaining dry ice can be carried out without special difficulties within the USSR at good factories in the center.

4. A considerable obstacle to the use of dry ice for cooling purposes is its high cost. In essence, dry ice must cost more than ordinary ice, even when machine-produced—

¹ See above.

… of the latter, since, in order to obtain dry ice, temperatures must be obtained that are considerably lower than those required for ordinary ice. Thus, the field of its application is naturally limited to that part where the use of ordinary ice does not produce a sufficient effect.

1. The compactness of dry ice as a source of cold offers prospects for its use as a refrigerant in the most vulnerable link of the refrigeration chain up to the present time—the delivery of perishable products from the place of production (a butter plant) or extraction (fish) to the refrigerator. The solution of this problem appears especially simple and attractive with the development of motor transport.

2. Proceeding from present-day conditions and the price ratio, the use of dry ice as a refrigerant in long-distance railway transportation of perishable products in some cases proves sufficiently profitable.

It would be incorrect to suppose that the author of the present article proposes the immediate and broad application of the indicated method of cooling with dry ice for the needs of refrigerated transport, and so forth. For this, of course, it is necessary to carry out a number of preliminary works, namely:

1. To resolve the questions of the preservation of dry ice under various conditions.

2. To develop rational methods of packing dry ice for its preservation and its use as a refrigerant.

3. To resolve the question of the possibility of regulating the cold when cooling with dry ice, since in many cases excessively intense cold is also undesirable.

4. To test the method of cooling with dry ice under the conditions of an experimental long-distance shipment, with careful measurement of all the conditions of the experiment, including the quality of the product being transported. Only after all the indicated work has been carried out can the question be considered sufficiently developed for the next issue to arise: its application on a broad, industrial scale.

The purpose of the present article is merely the desire to stimulate technical thought toward further development of this question.

Addendum.

When the present article had been written, the author obtained the report of Otto Hardung at the refrigeration conference in Berlin in January 1929.¹

The figures for the cost of dry ice given in this report are considerably lower than those cited above (~0.1 mark, i.e., ~5 kopecks per kg of dry ice), but at the same time the calculations assume very cheap current (~3 kopecks per kilowatt-hour), a considerably lower expenditure of labor, i.e., greater mechanization of production, and continuous production throughout the year. It is difficult to expect that all these favorable conditions could be applied directly in our country in the near future, especially under Siberian conditions. Nevertheless, these figures are extremely interesting, since they make it possible to consider that pro—

the savings figures we have calculated are far from the limit and may be exceeded to a considerable extent.

Of significant interest to us is the method of obtaining dry ice according to the process of F. Carba¹. The essence of this process consists in obtaining carbon-dioxide snow, under the pressure of the triple point, in the form of wet ice, which normally solidifies into high-density dry ice directly, without compression by a hydraulic press. This idea is extremely interesting, since it makes it possible to simplify and cheapen the production of dry ice considerably.

1. Carba—Bulletin (Bern) 2,1. ↩↩↩↩↩↩↩↩↩

2. The data were placed at the author’s disposal by the participant, Comrade Rybchenko. ↩↩

3. Although such calculations are not made in practice. ↩