“Soviet Planning and Spatial Efficiency”
PRODUCT MIX, RAW MATERIALS, AND COSTS
To judge an industry’s spatial efficiency we must know whether its production technology constrains location. Is the location of raw materials and fuel decisive? Are technological complementarities with other industries significant? Do scale economies dictate a highly concentrated pattern? Do real transport costs favor a more or less dispersed industry? In this chapter we will see that none of these factors exercises a significant constraint on the location of the cement industry. The analysis of Chapter IV makes it clear, however, that the widely dispersed production pattern and regional self-sufficiency which this locational freedom implies did not develop in the prewar planning era.
The industry’s product, hydraulic cements that are capable of hardening under water and resisting the prolonged, normally deleterious action of fresh or salt water, can be divided into three main groups: portland, portland-pozzolan, and portland-slag. In 1930 portland accounted for 92 percent of hydraulic cement output and the portland blends for 2 percent and 6 percent respectively; since then, the share of portland has fallen to about two thirds of the total, while the portland-slag and pozzolan shares have reached 8 percent and 26 percent respectively.1 In the United States industry, by contrast, over 99 percent of production is portland.
1. Portland Cement
a) Raw Materials Requirements. Portland cement is defined in the Soviet Union, much as it is in the United States, as the product resulting from the grinding of clinker, the predominating compounds in which are calcium silicates together with a sulphur compound which retards the set of the cement. The principal difference arises from the Russian tolerance, since March 1936, of up to 15 percent of hydraulic additive; the United States definition permits nothing to be added to the clinker except retarder and small amounts of certain other special-purpose substances (less than one percent). In this study when we wish to refer to the Soviet American-type Portland—that is, to a Portland containing no additive—we will speak of “pure Portland.”2
Depending on the character of the raw materials, portland cement is produced by one of two methods, the wet or the dry process of combining the raw materials prior to kilning. In either case the resulting kiln feed is calcined in a kiln, usually a rotary kiln, until incipient vitrification at a temperature of around 1500°
C. The resulting small knobs or clinker are ground to a specified fineness together with the sulphate setting retarder. After tests of its mechanical properties have been conducted, the cement powder is stored or shipped directly.
Typically the raw materials for pure portland cement are, for each ton: limestone, 1.5 tons; clay, 0.4 ton; and gypsum, 0.05 ton. The input coefficients reflect the somewhat inefficient prewar Soviet technology: dust losses of 15–20 percent were suffered in the kiln and grinding mill. Locational flexibility could be attained in some cases through the replacement of limestone by other limebearing materials such as chalk, shale, marl, marine shells, or cement rock, a limestone containing the cement oxides in almost exactly the proportions of portland and requiring only minor correction.
What of the fuel input? Before World War II coal accounted for about 80 percent of the industry’s fuel input, oil supplying the rest (today gas is the most important single fuel, providing about half of the industry’s caloric requirement). The coal input varied from 0.25 ton to 0.3 ton per ton of pure portland.
Table II-1 shows the amount of each material input, the quantity of each input remaining in the finished product, and the weight loss. Since the largest loss occurs in the limestone input, this becomes the most important locational constraint of all the materials. As will quickly become evident, however, limestone and other similar lime sources are widespread, as are the other materials, so that the raw materials do not exercise any serious constraint on location. This ordinarily allows production to locate near the market.
(b) Grading. In the early 1930’s there were three grades of Portland cement. In 1936 new standards were introduced providing for six grades, ranging from 200 to 6 00, the grade designation standing for the 28-day compressive strength in kilograms per square centimeter as measured by specified testing procedures. These same new standards introduced five grades for portland-pozzolan and six grades for Portland-slag.3 In addition, various special purpose cements, such as white Portland, early hardening Portland, oil-well cement, and others were introduced, although their production was very limited in the prewar period.
The many strength-grades of general-purpose Portland cement in the Soviet standards contrast sharply with the American single-grade practice, which requires the same minimum strength for all general-purpose Portland cement—that is, for all except certain cements not usually maintained in stock.4 The Soviet approach is probably a consequence of the industry’s success indicators (see Chapter III), and the reason for the American standard is to be found in the American preference for varying engineering designs (for example, the foundation cross-section) or modifying the design of the concrete mix to achieve different strengths.5 The single-grade approach also characterizes the industry of many European nations. Moreover, the fact that as early as 193 7 this concept was the rule in the Brazilian Portland cement industry6 suggests that it was not exigencies peculiar to an underdeveloped economy which conditioned Soviet standardization practice. The single-grade system undoubtedly simplifies construction by eliminating one variable from building plans, and it facilitates control over cement plant operations. Both these advantages should be especially helpful in a planned economy; and, as shown in Chapter V, such a grading system leads to sizable savings in the cost of delivery.
2. Portland-Pozzolan Cement
Pozzolan blend cements are cements based on a pozzolanic material, which, although having no cementing value itself, is capable of reacting with lime to form cementitious compounds. Pozzolan constituents may be blended with various bases; when they are mixed with portland cement, the result is called portland-pozzolan cement. Blended with lime, lower-strength lime-pozzolan is formed. Frequently, especially in official statistical tabulations, portland-pozzolan cement is called by the Russians simply “pozzolan cement” and this terminology will be adopted here.
In American portland-pozzolan cement the pozzolan is permitted to range between 15 and 50 percent of the weight of the finished product; in the Soviet Union this cement may contain between 20 and 50 percent of pozzolan (with less than 15 percent the product is still called portland cement, as noted above). In both countries it is permitted that the product be made either by grinding together the pazzolan and the clinker or by mixing thoroughly the separately ground constituents.
There are two advantages in using pozzolans in cement production: the first is that one grinding (the pre-kiln grinding that cement raw materials normally undergo) and the kilning of the pozzolan constituent can be dispensed with, thus lowering fuel and power costs. Average unit costs of production are therefore lower. For example, in 1957 they were reported to be 10.7 percent and 16.5 percent lower than portland cement produced at Briansk and Novorossiisk respectively, the two main pozzolan-producing areas.7 The second reason for using pozzolan cement is that it generally has a lower heat of hydration, which is an important property in mass structures and improves the hydraulic properties of a portland-base or lime-base cement. In this last regard Bogue observes that the superiority of pozzolans in resisting the attack of both fresh water and sea water has been “rather widely confirmed.”8 The Russians are quick to point to this advantage of pozzolan cement, but the apparent cost advantages are undoubtedly uppermost. For certain applications a third advantage of portland-pozzolan cement is that, since pozzolan reactions continue for a prolonged period, a lower 28-day strength portland-pozzolan may ultimately attain higher compressive strengths than would portland, so that if high early strengths are not necessary but later high strengths are, it may be cheaper to use the pozzolan.
It might be asked why pozzolan cement in the United States does not have the same cost advantages. Since portland-pozzolan has been used successfully in many American construction projects (the San Francisco piers of the Golden Gate and Oakland Bay Bridges, the Bonneville Dam),9 why is it not produced today in greater quantity in this country?
There are several possible explanations. The first is that more Portland-pozzolan may be produced than is reported by the Bureau of Mines in its annual compendium. The Bureau reports mill shipments, and it is possible that some cement leaves the mills as portland but actually is blended with pozzolan additives at the construction sites. A query about this possibility brought a reply from the Portland Cement Association that such blending could take place, but there would be no way to check into the extent of such practices. The Association itself does not encourage on-the-job blending, since it feels that cement can usually be purchased with the desired properties already incorporated in the product at lower cost.10
A second explanation is that the cost advantage may be more apparent than real. The reason is that since more portland-pozzolan than straight portland is required per unit of constant strength concrete, the production cost saving per ton of portland cement is offset by the extra cement required in construction (see next paragraph). If this is the case, the only justification for increasing pozzolan cement production could be that improvements in success indicators of producers could be achieved at lower cost thereby, since pozzolan and portland could be substituted in plan fulfillment, as will be apparent in the analysis of Chapter III.
That the cost saving was apparent rather than real may be concluded from the following considerations: in the 1930’s Professor V. G. Skramtaev, a well-known Soviet cement technologist, pointed out that 20 percent more portland-pozzolan than straight portland was required to produce a concrete of the same nominal strength. Skramtaev experimented with cements produced at Briansk, and he hastened to add that his conclusion should not be taken as general but that each mill might have its special relationship. Subsequently, it was suggested that the difference in input might even exceed 20 percent. More recently, Butt has estimated the difference as 15–20 percent. Meanwhile, the cost of producing portland-pozzolan, as indicated above, is only on the order of 10-16 percent less in recent years in two major cement-producing regions. It is unlikely that the relationship was vastly different in the 1930’s. Since the relative reduction in cost per ton of portland-pozzolan was about equal to the relative increment in input for concrete of constant 28-day strength, the total concrete production cost was approximately constant, and there would have been no national economic advantage to producing pozzolans, except insofar as their performance actually was superior in certain waterworks or other mass structures.11
3. Portland-Slag Cement
A waste product of the steel industry where limestone is used as a scavenger to draw off impurities from the ore melt, slag is produced in amounts of 0.60–0.75 ton per ton of pig iron. Two main uses of slag in cement production may be distinguished: (1) as an input to portland cement, substituting for the clay and a portion of the main lime-bearing input and (2) as a pozzolanic constituent in blended portland-slag cement. The first results in a Portland cement similar to that produced with the more usual ingredients. The second and far more important use of slag is as a pozzolanic ingredient in blended portland cement. The major difference between portland-slag and portland blends based on volcanic ash is that whereas the pozzolanic additive in the latter is limited to 50 percent, portland-slag may contain as much as 80 percent.12 Thus a ton of pure portland will yield 2.5 times as much portland-slag as portland-pozzolan, so that portland blends tend to be produced from slag rather than pozzolan. This accounts for the relatively rapid growth of portland-slag in the Soviet industry.
The production process is similar to that used in portland-pozzolan cement—that is, the slag is either ground with the portland clinker or thoroughly blended after granulation. Since use of the slag additive entails no weight loss, there is no compulsion to use it only at plants within the steel-producing regions; it could be efficiently shipped for use in plants elsewhere.
The relative ubiquity of cement-making raw materials means that industry location should be demand-oriented: production should tend to follow demand and there should be many producing points in many regions. This contrasts best, perhaps, with the steel industry, where location may be characterized as materials-oriented. The salient features of such a comparison for the Soviet cement and steel industries on the eve of World War II are presented in Table II-2.
As Table II-2 shows, the largest cement-producing region accounted for only 21.8 percent of total output while the largest iron-producing region turned out over 60 percent of national output. Nine of the thirteen regions each produced over 5 percent of the total cement production, while only four regions surpassed this share of total pig iron production. Notwithstanding the cement industry’s dispersion, however, transport costs in the 1930’s averaged around 50 percent of total delivered costs, and the annual average rail haul of cement ranged from 1000 to 1800 kilometers throughout most of the decade (Table IV-1). By contrast the average rail haul of all commodities in the decade ranged between 600 and 700 kilometers. And the cement haul even exceeded that of many commodities whose rare natural occurrence or substantial scale economies in production would be expected to result in long hauls—for examples, coal, less than 700 kilometers; iron and steel, 800–1000 kilometers; grain, around 700 kilometers; and petroleum and derivatives, 800-1200 kilometers.13
Some Measures of Locational Dispersion in the Steel and Hydraulic Cement Industries: 1940
Note: The 13 regions considered are those of Table IV-2 and include the whole of prewar USSR. Information for pig iron production is taken from M. Gardner Clark, The Economics of Soviet Steel (Harvard University Press, 1956), p. 238.Much the same picture is observed if steel ingot rather than pig iron is considered. The information on cement production is based on Table IV-2. If Baltic republics are included, the contrast is even greater, since they possessed facilities for the production of cement but none for steel.
There are two obvious causes for an irrationally long average haul for a commodity: incorrect location of the industry and inefficient distribution with irrational shipping patterns. Chapters IV and V examine these two possibilities, and the fault is seen to lie in poor location rather than inefficient distribution. As a preliminary to the analysis of those chapters we now survey briefly regional raw materials patterns and their effect on regional production cost variation.
1. Clinker Components
All through the 1930’s Portland cement was the major product, accounting for 90 percent of the industry’s total output at the beginning of the decade and about 60 percent in 1940. In addition, pure portland cement formed the base for portland-slag and portland-pozzolan. Therefore, around 70 percent of the industry’s total output in the late 1930’s passed through the portland clinker state (about 95 percent in the early years of the decade), and so we begin by focusing attention on the products that go to make up the clinker.
About two thirds of the weight of portland cement clinker is lime, which does not appear as such in nature but, rather, undergoes a loss of about 45 percent from the form in which it does appear—calcium carbonate in limestone, chalk, or other minerals. In addition dust losses in the kiln and grinding mills raise the input requirements to about 1.5, the total weight loss being about .86 ton per ton of pure portland. The source of the lime is therefore bound to exert the greatest locational pull on the industry. The distribution of lime sources is pictured in the map of Figure II-1.
The map shows the almost ubiquitous nature of calcium carbonate minerals. West of the Urals (longitude 70 ) no area is more than about 450 kilometers from a deposit, and only a few and less populous areas are as far as that. Most areas have their own deposits or are only a short distance away.
East of the Urals, it can be seen that most areas, at least most which in the early 1930’s were likely to be important consumers in the foreseeable future (those points along the Trans-Siberia-railway) lie close to deposits. Most of these were very close to sources of lime, and none was more than about 500 kilometers away. Thus in both east and west, the sources of lime were so well distributed that production could be undertaken near any point where demand should require it.
The map also shows in a general way the desirability of deposits since it distinguishes among marl, chalk, and limestone. Usually marl and chalk have important advantages over limestone. Marl may be soft, as it is near Amvrosievka in the Donets basin of eastern Ukraine resulting in low grinding costs. Or, if it is somewhat harder, as it was in the Novorossiisk area of the Northern Caucasus, it may be so like Portland cement in oxide composition that it can be roasted without expensive pre-kiln grinding and mixing. These characteristics permitted these two areas to produce the cheapest portland cement in the country both prior to the Revolution and in the 1920’s. Novorossiisk marls began to be exhausted around 1930, however, and less desirable deposits not adjacent to plant sites had to be assimilated by the early 1930’s, necessitating some grinding and mixing costs.14
DISTRIBUTION OF LIME-CONTAINING NATURAL RAW MATERIALS AND MAJOR CENTERS OF CEMENT PRODUCTION-1936
SOURCE: RAW MATERIALS LOCATION FROM I S. LUR’E, “O SPOSOBAKH PROIZVODSTVA PORTLANDTSEMENTNOGO KLINKERA,” Ts., 1958, No. 2. PRODUCTION LOCATIONS FROM VARIOUS SOURCES.
Next to marl, chalk is the best source of lime. It is soft and easier to grind than limestone. The chalk deposits shown on the lower Volga, in the European west, in Belorussia, and in the European southwest were the basis for portland cement production at Vol’sk, Krichev, Briansk, and Belgorod.
The most notable kinds of variation among limestones are hardness, lie of deposit, and magnesia contents; these variations are not shown on the map. For example, limestones of the Podol’sk plant near Moscow are extremely hard, a fact noted as early as 1880, when production was undertaken there.15 The limestones in the Urals and in Siberia are also hard and require either a higher expenditure for grinding or a double pass through the kilns with attendant higher fuel costs.16 High magnesia content usually renders limestone unsuitable for portland cement production, although the development in recent years of magnesian portland, containing up to 10 percent magnesia, makes possible the use of impure limestones. Finally, the lie and quality of deposit and the ease of accessibility are important features conditioning the usability of a limestone formation.
It is difficult to determine the actual variation of quarry costs. Naumova presents data for 1959 for about twenty plants, including some in the chalk or marl areas (where, owing to softness and, usually, greater proximity of deposits, quarrying costs are lower). From her data it is evident that, as mentioned above, Podol’sk, Siberian, and Ural quarrying costs tend to be high, although the higher costs in the East must reflect the generally higher costs of eastern operations (a few plants in the East have relatively low materials acquisition costs).
Many of the plants listed by Naumova were not in operation in the 1930’s, but there is no reason to doubt that the 1959 relative cost range would also have reflected relative materials costs in the 1930’s for the areas in which the plants were eventually located. Yet although the variation in cost of materials was very wide, total materials costs were only a minor expenditure in cement production. In 1932 they amounted to only 20 percent of total production costs, and this is believed to include the wear of steel balls in the grinding mills. By contrast, in the 1930’s the freight cost for transporting the product to market averaged 100 percent of the total production costs, thus emphasizing that the market, rather than superior materials, should have been the major locational determinant.
The other important clinker input is clay. The amount required is relatively small, however, and the weight loss is low (about 0.1 ton per ton); therefore, if local limestone is suitable and demand is high, it will always pay to undertake local production even if clay has to be imported to the area, since it will be carried in tonnages only 40 percent the weight of the cement that would otherwise have to be imported and, almost certainly, over much shorter distances, since clay would most probably be available between the point of potential export to the region and any point where it is proposed to undertake production. A saving in the cost of transport would still be achieved even if coal had to be imported to the region. A further advantage is that much less care is required in shipping clay than cement. In any event, economically useful clays are available in most areas, and certain substitutes are sometimes used (for example, shale in Karaganda).17
2. Other Raw Materials
Besides the clinker components, the cement industry required coal, pozzolan additives, blast furnace slag, and gypsum to produce the output mix of the 1930’s. How did these materials influence location?
Less than 0.3 ton of coal was required to produce one ton of Portland cement clinker in the 1930’s, making coal the third largest input in terms of weight. Almost the whole of this weight is lost, only about 10 percent ash combining in the roast. But in spite of the very high relative weight loss, the absolute weight loss was two thirds less than the weight loss of limestone. Therefore, when both materials are not located together, it is rational to ship the coal rather than the limestone. However, coal itself was widely distributed, occurring in almost all of the small individual union republics and in many of the important regions of the RSFSR.18 At worst, then, coal would have to be shipped only short distances to permit production if a local market had a limestone base and enough demand. Moreover, the further possibility of substituting oil or gas for coal weakened still further the constraint of this input.
Blast furnace slag availability naturally attracted cement production since Portland-slag cost less to produce than Portland. Since there are indications that in plan-fulfillment the former could be substituted for the latter on a one-for-one basis (see Chapter III), there was naturally an incentive to produce portland-slag. However, since there is no weight loss in the slag input, there is no reason for not shipping slags to be used as an input in cement production in other consuming regions, rather than producing in the regions originating slag and shipping the final product from the latter to the former. Not only would there be savings in shipping weight, 20–25 percent, but savings in packing cost or reductions in dust loss en route would also be gained. Much the same principles apply to other pozzolan additives, such as volcanic ash.
The last material input of consequence is gypsum, used as a setting retarder to maintain plasticity in concrete until emplacement. The amount actually used in any instance depends on the mineralogical composition of the cement. The state standard indirectly limits its use to a maximum of about .05 ton per ton of cement. An average industry-wide input per ton of around .04-.05 ton is therefore suggested, and this is not enough to exert a major influence on location.
If the wet process is used, water is needed. However, the requirement—less than 500 gallons per ton—is very little compared to many other major industries. Viewed in this light, the cement industry’s need for water, which could be obviated by recourse to the dry process if raw materials are suitable, is small indeed and would not be a serious impediment to location.
C. REGIONAL VARIATION OF PRODUCTION COSTS
We have just seen that raw materials and fuel were widely accessible, although the quality of the resource base did vary somewhat among regions. This suggests some limited regional variation in factor cost. Unfortunately, complete cost data are not available in the industry’s literature of the period. There are many reasons to believe, however, that regional mill prices, which were the basis for cement pricing before 1936 and were published, do accurately mirror regional production costs. This view is supported by many considerations. First, the relative simplicity of the production process means that difficult joint cost allocation problems do not arise. Next, a few published cost comparisons for the period, as well as available regional cost information for the Tsarist period,19 are consistent with the hypothesized factor cost-price equality. What is known of the resource base of the various regions is also consistent with this thesis: the cheap Ukrainian production was based on good coal and soft marls; Leningrad production was based on poor limestone and imported coal; Volga production was based on an excellent lime base. The price quotation for Sverdlovsk may be exceptional in not reflecting true cost, owing to the receipt of a subsidy in the form of a preferential rate on coal haulage from West Siberia.20 Making allowance for this subsidy would raise the cost in 1932–35 by 4 rubles a ton—that is, to 33 rubles. The high prices in the Far East and Central Asia reflect the higher costs of eastern operations in general.
Table II-3 shows the mill prices of cement in 13 producing regions in 1929/30 and 1932–35. The considerations summarized in the last paragraph imply that we can safely use these price data as cost inputs for our analysis of the efficiency of alternative location arrangements in Chapter IV, and we shall henceforth treat them as costs.
There is a concentration of plants producing at a cost of between 25.80 rubles and 29.05 rubles after 1932. The 1929/30 price list is also notable for the narrow range of most producing regions. Only three regions—the Far East, Central Asia/Kazakh SSR, and the Leningrad region—are marked by costs appreciably higher than the average. All the others could produce within 20 percent of the average, which was 28.0 rubles in the 1929/30 price list and 27.8 in the 1932–35 list. It might be observed in passing that a narrow cost spread is also observed in the American portland cement industry in the 1930’s.21
Although three regions show costs on the order of 30–70 percent greater than the average of the lower cost regions, regional production cost differentials are not critical. In the first place, if production were not to be undertaken in these three regions, delivered costs would soar, since cement would have to be shipped 2500–11000 kilometers at a cost of 50–200 rubles per ton, and this would more than offset the production cost advantage of 15–25 rubles of the low-cost regions. But even the production cost variation pictured in Table II-3 is much more limited than would be observed in an industry such as steel, where, except for a few major producing regions, both iron ore and coal would have to be imported if local production were insisted upon. This, then, is the frame of reference within which the statement that the industry has few technological constraints or that production costs show little regional variation must be understood; delivered costs would be much higher in most regions if cement had to be imported from the lowest cost-producing regions rather than produced locally. More precise comparisons are made in Chapter IV.
Regional Mill Prices of Portland Cement in Bulk, 1929/30 and 1932–35
Source: Prices for the 1929/30 fiscal year are from “Tseny 1929/30,” SM (1930), No. 3, p. 112. Prices given here reflect regional averages where more than one plant price is quoted in the source. Prices are converted to bulk prices when only the barreled price is given in the source. Most of the plants to which the prices apply are identified completely. For example, the Leningrad Vorovskii plant is listed under Leningrad. Other plants, such as those at Vol’sk, are given by plant name only, but information elsewhere helps to place their location. 1932–35 prices are from Spravochnik 1935, p. 26. The prices quoted here are those for the middle grade (00) of Portland cement. These prices went into effect in 1932.
D. ECONOMIES OF SCALE AND TRANSPORTATION COSTS
In the prewar period the optimal scale of cement plant in most regions was between 75,000 and 120,000 tons. Larger plants, several of which were planned during the decade, were unlikely to be economical even though they could attain some small savings in production costs. Their irrationality was to be found in their long construction periods and the excess burden they imposed on the transport network. This conclusion is based on analysis of the industry’s technological lines and Soviet cost data for the period and is reinforced by studies of the American industry which show only slightly decreasing—or even increasing—costs at moderate plant size.
1. The Production Line
The portland cement production line is usually based on the rotary kiln, a development of the late nineteenth century. The main components of the line are the crushing and grinding department, the slurry basin which mixes the ground raw materials, the rotary kiln, and the clinker grinding mill. (The dry process bypasses the basin.) The sizes of all the units are variable so that kiln and clinker grinding-mills can be enlarged together. Technology for the 1930’s was based on one clinker grinding-mill and two raw materials grinding-mills per kiln. A single basin serves all production lines.22
Clinker production is a continuous process, the kilns being idle only for repair and refurbishing. The rotation of the kiln keeps the raw materials from adhering to the inner walls and improves product quality, a fact of great importance in the kiln’s rapid proliferation and displacement of the older shaft kiln technology, which will be described briefly below. The rotary kiln requires less labor per ton of clinker than the shaft kiln.
The length of the rotary kiln is the most important parameter of portland cement technology and is the most important influence on capacity. If a larger plant is wanted, a longer kiln is planned and the capacities of other components increased correspondingly. But the prewar era was still a period of relatively short kilns in the USSR. The longest were 108 meters, and almost all kilns built after 1940 were between 118 and 150 meters.23 This meant that certain capacities would require two smaller kilns rather than one very large and one small. For example, if the largest kiln has a capacity of 115,000 tons a year and a plant capacity of 150,000 tons is planned, the best solution is to build a plant with two kilns of 75,000 each. To produce 230,000 tons a year a plant should be designed with two kilns of 115,000 tons apiece. These two kiln capacities correspond to lengths of 70 and 111 meters respectively.24
The longer the kiln, the lower the unit production costs; capital costs and labor costs are lower and there is probably some small saving in power costs per unit of product. To double capacity after the maximum size of kiln is reached, the plant designer must add a second maximum-sized production line. The only source of decreasing unit costs in this situation would be the economies associated with general overhead (laboratory, marketing, administration), quarrying, and the slurry basin, and these would be relatively low compared with the more directly variable clinker-production costs. Thus a long-run average-cost function is suggested in which costs decline rapidly to 116,000 tons, then fall a little further at 232,000 tons, and so on.
2. Soviet Cost Analyses of the 1930’s
Two Soviet cost analyses for the thirties enable us to assess the behavior of cost with respect to plant size. The first provides information on capital costs of new plants projected during the First Five Year Plan, and the second relates labor input to plant size. Since other production elements do not show declining unit costs (or have only slightly decreasing costs) with increases in plant size, these items comprise almost the whole of the cost category from which scale economies would arise. These elements together accounted for 50 percent of the industry’s production costs in 1932.25
a) Capital Costs. A 1931 study contains estimates of the capital costs of twelve Portland cement and three Portland-slag plants projected for construction as of 1931.26 The unit capital costs are shown in column 1 of Table II-4.
The cost per ton of cement capacity is at first constant at around 71.50 rubles for plants of 116, 250–232,000 tons capacity, a large plant for that period. No plants were projected between that size and 465,000 tons. Costs are given in the source for six plants of the latter size. Their average cost per ton (and they are all closely grouped around the average) is 51.70 rubles, a decline of around 28 percent from the 233,000-ton plant. If the life of the plant is estimated at 30 years—and this is conservative—the saving per ton of output over the life of the largest plant is 1.10 rubles, as compared with the smallest plant. This saving amounts to only about 4 percent of the average production cost, which was 26.70 rubles in the early 1930’s.27 Few regions could support the largest plants—indeed, in the decade only five regions were consuming more than 250,000 tons annually. And we will see shortly that real transportation costs were too high to justify concentration and distribution from large plants.
b. Labor Costs. In the study of labor costs M. Shmukker assumed crews of size appropriate to handle the various units in production lines of different capacities.28 He considered plants ranging in size from 93,000 to 465,000 tons.
The essentials of Shmukker’s analysis are shown in columns 3 and 4 of Table II-4. The labor inputs are given only in physical terms, however, and must be converted to rubles. To convert the physical input series into value terms, we start by noting that labor costs in the late 1920’s averaged 41 percent of total cost, or 10.9 rubles per ton.29 Most of the plants of the time were small and old (pre-World War I), however, and were operating short of capacity. These factors suggest that actual total production costs and labor costs were running higher than they would have been in larger and efficiently operating new plants. Accordingly, we reduce the total cost estimate to 22.5 rubles and assume a 30-percent labor component, or 7.5 rubles per ton for the 93,000-ton plant.
The labor cost analysis demonstrates very large saving in labor input. As plant size increases from 93,000 tons a year to 465,000 tons, labor costs fall by two thirds. Since the capital costs decline less rapidly, however, the total saving in production cost does not fall proportionally, a decrease of 58 percent being observed in moving from the smallest to the largest plant projected, as shown in column 7.
3. Other Cost Studies
Other cost analyses were studied as a check on the foregoing evaluation of scale economies in the prewar Soviet cement industry. These include studies of the American industry by the Temporary National Economic Committee and by Joe Bain. The TNEC data, which refer to 1929, show the optimal American cement plant to be of medium size (170,000–340,000 metric tons per year). The thirteen lowest cost plants in this study fell within this range.30 Larger plants had higher unit production costs.
It should be remarked that the higher unit cost of the very large plant in the American industry does not reflect higher costs brought about by extension of the market and consequent longer shipping radius. Production costs themselves, it appears, begin to rise in the American industry. Even the efficient American plant (that is, the medium-size plants of 170,000–340,000 ton capacity) would most probably have shown higher total delivered costs and would have been undesirable if distances and market sizes had been comparable to those in the Soviet Union. Thus, while American experience does not tell us precisely what the efficient size would have been in the Soviet economy, the principle of efficiency of medium-size plant is consistent with what we found there.
Sources: Capital costs from A. Greiman, “O postroechnoi stoimosti novykh tsementnykh zavodov,” SM (1931), No. 2–3. Those marked with an asterisk are interpolated into Greiman’s data. Others are averages for several plants of projected size. Greiman’s data are given in terms of barrels and are converted here to metric tons at the rate of 155 kilograms per barrel. This conversion is also applied elsewhere.
That the portland cement barrel traditionally weighed 155 kilograms is stated in various places. Sometime in the 1920’s a heavier barrel came into use in the Novorossiisk area (170 kilograms), as an economy move. It is believed, however, that, even in this region, when plant capacities or production figures are given in barrels, 155-kilogram barrels are meant. See I Pavlenko, “Sostoianie ratsionalizatorskikh meropriiatii Novorostsementa,” SM (1930), No. 7–8, p. 170.
Labor inputs (physical terms) from M. Shmukker, “Normal’nye shtaty dlia proektiruemykh tsemzavodov,” SM (1930), No. 2. For the derivation of monetary labor and transportation costs see text.
The conclusion that the industry is not one of continuing economies of scale is also reached by Professor Bain. He bases his conclusion on the TNEC study and, in more recent years, on other empirical analyses.31
4. The Scale-Transport Trade-Off
With the scale economies shown in the table we can calculate the incremental length of haul that could be supported by the larger plants. The trade-off calculation is based on a cost of 1.07 kopecks per ton-kilometer as the cost of hauling cement by rail in 1930,32 and the result is given in Column 9 of Table II-4 and shown graphically in Figure II-2. Whenever the distance added to a plant’s average haul is less than the distance given by the ordinate of the curve, it would pay to build a plant of the size in question. For example, suppose that to serve a given large market area, a 279,000 ton plant with an expected average haul of 750 kilometers could be built. Suppose, alternatively, that the same large market could be served by three 93,000 ton plants, each with an expected average haul of 400 kilometers. In this case, the large plant should be built, since the additional haul required would be only 350 kilometers, while the ordinate at 279,000 tons is 399 kilometers. If, on the other hand, the three smaller plants were each expected to have an average haul of 300 kilometers, the smaller plant variant should be adopted, since the incremental distance for the large plant variant would be 450 kilometers—that is, 51 kilometers more than the maximum permitted by the scale economies of the larger plant. It is evident from this example and from knowledge of the production process that the total scale economies in the delivered cost of cement are not likely to be great.
There is, finally, another cost associated with transportation of cement: the dust loss en route. The longer the distance, the greater the loss. At some point the loss is so great that it is cheaper to pack the cement in barrels or bags, adding about 10 percent to the mill cost.33
One last argument against the very large plant is the protracted construction period involved—eight years and more, as it developed, in the 1930’s. For example, the Gigant plant, a giant 465,000 ton plant near Moscow, originally had a construction schedule stretching from 1930 to 1937.34 It is believed that this plant was still not in full operation as late as 1940. The 350,000 ton complex at Spassk in the Far East was originally scheduled for completion in two sections, the first in 1934 and the second in 1937 (construction began in 1931).35 The Spassk plant in the Far East was producing only 234,000 tons in 1940, however. The foregoing examples do not necessarily imply locational irrationality. The Moscow region was a major consumer, and the production economies of Gigant need not have been lost to increasing transport costs. The Far East, too, could easily use a plant the size of Spassk (it had no others), and Spassk was close to the consumption centers at Vladivostok and Khabarovsk. The two-stage construction schedule was a further testament to rational behavior on the part of planners. All that is intended here is to emphasize that big plants take a long time to build, and if cement is needed two years from now, it is small consolation to know that after a few more years cement will be provided at somewhat lowered cost. And it is interesting to find that of the six 465,000 ton plants projected in 1930, cited earlier, only the Gigant plant was actually undertaken.
Incremental Average Haul Permitted for Plant of Given Size as Compared with 93,000-ton Plant
Source: Table II-4
An appreciation of the colossal size of the 465,000 ton plant may best be derived by reference to the advice of V. N. lung (perhaps the most eminent Soviet scientific expert on cement), and S. P. Preobrazhenskii, given on their return from a tour of the American industry in 1928. Their conclusion—and it was a daring innovation—was that a plant of a capacity of 230,000 tons should be built in the Moscow region36—that is, a plant one half the size of the Gigant plant which was projected in 1930.
One last consideration weighing against large plants is the capacity of the quarry, something frequently overlooked in discussions of scale and in actual construction projects. For example, in 1956 two additional kilns went into operation at the Belgorod plant. With the new kiln capacity, the quarry life was re-estimated at 18 years.37 With an amortization period of thirty years for the plant (and probably a much longer actual operating life), it is not hard to see that any cost calculations that were relevant at the start would be unrealistic early in the life of this equipment.
5. The Shaft Kiln
The traditional oven in the cement industry in the nineteenth century, the shaft kiln, was displaced in the twentieth century in the cement industry of modern industrial nations by the fuel-intensive, labor-saving rotary kiln. As might be expected, the substitution was most rapid in the United States, where the cement industry was heavily concentrated in or near the great coal regions of Pennsylvania and labor was in short supply. The first rotary kiln in Russia appeared in 1908; by 1912 there were forty-four rotary kilns out of a total of 239 kilns in use or in construction.38 In 1910 the first of a series of measures was put into practice to mechanize the shaft kiln through mechanical loading of the raw materials and unloading of the clinker. Product quality improved as well.39 These improvements caused a modest renaissance in the use of the shaft kiln in various nations. In the Soviet Union, ten of the so-called “automatic shaft kilns” were in place by 1930 and fourteen more were scheduled for addition in 1931. The kiln’s fuel requirement is about 25–50 percent less than that of the rotary kiln. Such a kiln would have a capacity of 20,000–40,000 tons a year, depending on the number of shifts operated. Unfortunately, comparative cost data for the 1930’s do not exist, but an analysis in 1947 indicates that unit production costs of the shaft kiln are not substantially higher than those of cement produced by rotary kiln; its expected electric power input is somewhat greater than the industry-wide average (103 kwh versus 89), its expected labor productivity, about 25 percent less (238 tons per man-year versus 318).40 However, since fuel consumption with the shaft kiln would average 25–50 percent less, the net result is a cost advantage of only about 10 percent for production by rotary kiln in a plant with a capacity of 100,000 tons. These relationships probably held for the 1930’s as well, so that the output of the shaft kiln at that time should have cost only a little more than that of the rotary kiln, and it is possible that production based on it should have been extended further, depending on whether there were pockets of demand of 20,000–40,000 tons per year isolated from areas already producing. Certain regions where the shaft kiln would have been the efficient solution are given in Chapter IV. It is interesting to observe at this point, however, that the new “ automatic shaft kiln” was not used in outlying areas at all. Rather, the twenty-four new shaft kilns built or scheduled for construction by 1931 were all at large cement plants, eighteen of them at Amrosievka and Novorossiisk.41 The reason for these locations is to be found in the raw-materials base, marl being well suited to the shaft kiln. The other six shaft kilns were planned for the Podgornoe plant in Voronezh.
We use cookies to analyze our traffic. Please decide if you are willing to accept cookies from our website. You can change this setting anytime in Privacy Settings.