Campaign Time Pattern, Sortie Rate, and Base Location

The exact time development of campaigns has not figured in most of RAND's formal campaign analyses, but so far as the Air Force is concerned, it is clear that it gives such considerations great weight. For example, General Vandenberg before the Armed Forces and Foreign Relation Committees of the Senate last spring, talking of the Strategic Air Command, stated:

"Obviously if you have to refuel two or three times in the air you get two or three missions per month out of your airplane instead of 15 or 20. Therefore your Air Force, in order to do that efficiently, would have to be five or six times the size it would be if we had bases that were more nearly appropriate. Additionally it would take longer to have the full impact of the air war upon whoever we fought."

With such considerations in mind we have presented measures of systems cost per sortie as well as systems cost per bomber.[2] The first are intended as a preliminary indication (in advance of a complete campaign calculation taking into account such important factors as attrition of aircraft over enemy defenses) of the cost of maintaining a given level of attack from various locations. The second[3] indicates (again neglecting attrition and other important factors measurable only by a campaign calculation) comparative costs during certain campaigns in which differences in carrier utilization over time, in missions flown per carrier, are insignificant. We cited two such campaigns: (a) the single-strike Sunday punch, and (b) a campaign conducted wholly with surface-to-surface missiles.

There is, of course, no doubt whatsoever that neither of these ratios measures exactly what we want. The costs of target destruction in campaigns of fixed or indefinite length, involving constant or variable rates of destruction, are a much closer approximation. For this reason D-1114 stated[4] the importance of extending the analysis to include the effects of attrition over enemy defenses, etc. It has been suggested however, that the systems-cost-per-sortie measure is inappropriate because campaign time and the rate of destruction are unimportant. This objection to the cost-per-sortie ratio is quite distinct from a recognition of its limitations from a campaign point of view. It would apply just as well to a measure of costs obtained through actual campaign calculations for a variety of fixed-length or continuing campaigns. We regard campaign analyses as an important next step in the base study; but not one that is magical in its research potentiality. The assumptions fed into the campaign calculations as to the targets, intervals between strikes, Soviet defense, etc., will decide the relevance of their results. To be specific, the view that the time pattern of the campaign has no importance appears to us quite doubtful. We would like therefore not only (I) to fix more exactly the role of the cost per sortie measure in D-1114, but also, (II) to make some comments on the more general question of the importance of campaign time and the time rate of attack.

I. Costs per Bomber and Costs per Sortie in D-1114

Now there are two sorts of comparative cost analyses in D-1114 which are presented in the form of relative costs per sortie. (a) In one, illustrated by the air-vs-ground refueling comparison in Section 12, the use of the sortie rate does not affect the result at all. (b) In the other, exemplified by the air-refueled primary base location comparisons of Section 13, the sortie rate effect contributes to the difference shown but these differences remain substantial even if we neglect the sortie rate and the problem of maintaining a given level of effort over time, and deal only with the cost per bombing aircraft. (c) A third sort of comparison mentioned in D-1114, namely a ground-refueled primary base location comparison, might be seriously affected, but we have not as yet undertaken the comparison.

(a) Air-vs-ground refueling costs. In this comparison, which shows a very considerable economy in ground refueling, Limestone is the starting point for the air-refueled as well as the ground-refueled missions. The differences in mission time, mission length, and sortie rate, such as they are, have been weighted against ground refueling by the assumptions that ground refueling involves a dog-leg but air-refueled flight is in a straight line and uses up no flying hours in the loiter period waiting for rendezvous. However, the differences are very small. The sortie rate per month, using the B-47, against Target I, Leningrad, for the ground refueled case is 6.2 as distinct from 6.3 for the air refueled case. To illustrate the effect of this slight difference, cost per sortie for the ground refueled system, assuming one refueling base per bomber wing, and ten-year base amortization, is $2.3 million. For the air-refueled system, assuming one tanker per refueling, $8.6 million. The costs per bomber, that is, the cost of the bomber plus its share of tanker or refueling base costs, are $17.2 million and $63.2 million respectively. The effect, in short, of substituting costs per bomber for costs per sortie are negligible.

The comparison presents a total of systems costs including both combat aircraft in active status on the one hand, and refueling aircraft or refueling bases on the other. This depends both on the relative costs of the two refueling methods which is what interests us, and the extent to which refueling is employed. The proportion of refueling to combat aircraft needed for the active campaign as distinct from those needed for the stock of reserves for replacing attrited bombers within any air or ground refueled system, depends on the policy for reserving or using combat aircraft during the campaign, as well as on the range to target. If we send all usable bombers out on every strike then the maximum in refueling resources are needed at the start. Refueling has to take care of the peak bomber traffic. If we reserve bombers as replacements for later strikes, the proportional importance of refueling in total bombing system costs will be less. Nonetheless, for the refueling part of the job, the relative costs of alternative methods is unchanged, and no matter what policy is followed, air refueled SAC missions from the ZI against the total industrial target system entail several billion dollars. Therefore the comparative costs of alternative refueling methods are of substantial interest. In our illustration above (one tanker wing or refueling base per refueling) the ratio of the cost of purchasing and operating the tankers as compared to purchasing and operating the refueling bases is 24 to 1.[5]

The air-vs-ground refueling comparison is one of two principal illustrations used by D-1114 which is primarily concerned with an over-all analysis of the systematic connection of factors in base location. These two illustrations were chosen because they display differences gross enough to survive the probable alterations in systems costs to be attained by completer study. One part of a completer study would take into account the effects of attrition during appropriately set up campaigns. We do not imply that the differences in favor of the apparently superior systems will be the same in quantity, say 24 to 1 or 10 to 1; just that it appears to us that they will remain and remain distinct. We believe the chances are particularly strong for the air-ground comparison. In fact in D-1114 we have listed a large number of factors deliberately left out of account such as the problem of rendezvous in air refueling which would tend when considered to increase the difference shown in favor of ground refueling.

(b) Air refueled primary base location. Section 13 of D-1114 shows very substantial differences among the costs per sortie from various primary base locations. Sortie costs, to the entire industrial target system from a ZI base complex are more than ten times those from a "least-sortie-cost" overseas base complex. And against specific target areas there are similarly very wide differences. These differences are affected by variations in the sortie rate, but they are by no means completely determined by them.

D-1114 states[6] that the more than ten to one difference displayed between the ZI and overseas base costs is a product of three factors of which the sortie rate variation is only one: The other two are the differences in the number of refuelings required, and the difference in the cost of the B-47 and the larger B-52 used in the ZI operation. The sortie rate from the ZI in our example comes to some 37 percent of the sortie rate from the overseas base complex. But, besides this, an average of 1.5 refuelings are required from the ZI and none from overseas; and the bomber is roughly twice as costly. As a result costs (bomber plus required tankers) per bomber from the ZI (neglecting sortie rate difference) are still better than four times those from the least cost overseas complex.[7]

The relative ranking of bases with respect to specific target groups is not substantially changed.

Now, as we stated in D-1114, while we feel there is a distinct superiority displayed by systems involving at least refueling bases overseas, we do not believe that there are sufficient grounds at this stage of the investigation for choice among the types of overseas base systems outlined in Section 11 of D-1114 (e.g., overseas primary versus overseas refueling bases, overseas peace-and-war versus overseas wartime only). Uncertainties in the estimates of primary base vulnerability, open questions concerning the importance of campaign time against various classes of targets all effect the outcome of analysis.

However, it might be worth anticipating what the effect of neglecting time constraints and sortie rate differences might be on a comparison of ground refueled ZI bases with an overseas primary base complex. A very rough calculation of the costs per sortie of ZI primary base systems supplemented by a hypothetical overseas ground system (calculated merely by assuming two overseas refueling bases per ZI wing) shows sortie costs of about two and a half times the overseas primary base complex when the B-47 is used both in the ZI and overseas, and a little less than five times the overseas case when the B-52 is used in the ZI and the B-47 overseas. (This compares with the ten to one difference for the air refueled B-52 ZI based system versus the B-47 overseas.) If we assume there are not time or rate constraints whatsoever, then costs for the ground refueled ZI based B-52 are less than double the costs for the overseas B-47 system. The costs for the ground refueled ZI based B-47 are substantially the same as for the overseas system. This last point does not mean that the two systems mentioned are equally good, (or equally costly in fact and that the choice must be made on other grounds), but just that at this stage, for the B-47 case, if sortie rate effects are neglected, we have no clear economic grounds for choice.

If campaign time patterns are unimportant, then other alternative base systems would also be affected favorably. For example, in a system basing aircraft in the ZI in peace and overseas in war, the air lift requirements would be drastically cut. The margin of uncertainty in our estimates of the system costs and the doubts as to the magnitude of primary base defense cost here are large enough to throw the result in any of several ways. For this reason, D-1114 postponed these comparisons.[8]

Table II summarizes points (a), (b) and (c). For the first point it shows ratios of cost per bomber as well as cost per sortie. For the second and third, it shows the ratio of ZI to overseas systems costs, with and without campaign time or rate constraints, for both air-refueled and ground-refueled systems. Three levels of active defense are shown.

II.Campaign Time Patterns

(b) The sortie rate is unimportant

1. if there are no time constraints whatsoever.

2. in the case of the most extreme time constraint where campaign time is limited to one day as in the "pure" Sunday punch theory.

3. where there are no time limits, but these are relaxed enough to be inoperative so far as differences in periods of air craft utilization are concerned.

While neither of the two principal comparisons made in D-1114 hangs on the sortie rate variations among base locations, this does not mean that we do not consider these variations important. The sortie rate may be a critical factor in continuing campaigns involving a constant rate of destruction (Bodenhorn), in campaigns involving a schedule of destruction at an increasing rate (Goldhamer) as well as those involving a high early concentration of bombing (modified Sunday punch) and in various other fixed-length campaigns.

Consider in turn each of the conditions listed as making variations in sortie capability of no importance.

1. No time constraints. Suppose a closed system of targets, with no recuperability and a fixed initial stock of aircraft and bombs to be used up in the course of destroying the targets. So far as cost criteria are concerned, in this case, the peacetime purchase and maintenance costs of this initial stock are in general determinative of weapon choice. Such a model is perhaps least plausible, or consistently applied, if there is no time constraint whatsoever. It has under this circumstance a number of absurd consequences, only one of which we shall mention.

   If we are to take seriously the assumption that time is a matter of indifference entirely, then it is not very clear why we bother to have a large initial stockpile. For all the missions for which time is irrelevant, assuming there are any, we could slough off the considerable burden of maintaining a peacetime force and just depend upon production after the outbreak of war to provide the requisite number of planes to conduct a given strike and then wait for further production to bring the stock up for the next strike. One need not go to the extreme of waiting till the outbreak of war to send out performance specs. One might make concessions of a moderate sort to the exigency of time and have pilot production in time of peace. This would permit a comparatively rapid rise in time of war, perhaps enough to contribute substantially to aircraft requirements of a campaign lasting a year or two. However, it is clear that a wholly stockpile war is not rational for this case.

2. The pure Sunday punch theory. That it is important to compress the entire strategic air campaign into a few hours, in Brodie's "hoss opera" phrase -- to shoot from the hip and empty the magazine -- appears to be the polar opposite of the preceding theory which says that time is of no importance at all. They are distinct, but, in fact, have some essentials in common. Both views assume (a) a closed target system, (b) that we will actually have the force requirements to eliminate it with a minimal number of strikes -- in the case of the Sunday punch, just one, and 9c) that elimination is final, recuperation may be neglected.

   It is not necessary to deal with Sunday punch theory at length here. Brodie's vigorous comment on the subject indicates some of the weaknesses of several variants of the view. In fact, it is only the weakest, most extreme versions, the pure single strike theory, that make sortie rate irrelevant: if there is only one strike, then of course the possibility of reusing an aircraft on another strike is not very interesting. As soon as realities force an extension of time, as Sunday extends to a month or more, as it has in Air Force planning today, then the theory thins down to advocacy of a time pattern of attacks involving high early concentrations. And for such patterns (but not only for such patterns), locational variations in sortie rates are extremely important. Campaign time is of great interest to the Air Force.

   The importance that is assigned to time in present Air Force thinking is illustrated by the SAC mobility plan. This plan is a long way from the Sunday punch. However, it does involve the expenditure of a considerable amount of resources in order to provide the capability of transferring SAC wings from primary bases in the Zone of the Interior to wartime primary bases overseas and to do this quickly. If time were not a consideration a more majestic pace of transfer would certainly reduce the air lift expenditures significantly. On the other hand, if time were not essential, then the plan for putting primary bases overseas in wartime would well be subject to question. It would still, as our study suggests, offer large advantages as against air refueled sorties from the ZI. But it is doubtful that it would offer any advantages over ground refueled sorties.

3. Campaigns with non-critical time patterns. A stronger case might be made for the irrelevance of such time variables as the sortie rate than is presented either in the no-time-limits theory or in the pure Sunday punch view. This argument might run as follows:

   "There is a fixed number of strategic targets and a desirable total time limit on the campaign to destroy them. Considera tions of enemy defense levels, etc., determine the optimum number of targets to attack on any one strike and therefore (considering various parameters determining effectiveness) the optimum number of strikes. This number can in fact all be mounted with the forces we will have available from the remotest (or lowest sortie rate) bases within the desirable total campaign period. In this case the only restriction on the intervals be tween strikes is the average length implicit in the over-all time limit and the total number of strikes needed; the shortest desir able interval between strikes is no less than the mission time from the remotest or lowest sortie rate base plus the period of maintenance for the force needed for the next strike. (Main tenance here includes both damaged and undamaged planes.) There fore the time pattern is not critical."

   We have made this argument as strongly as we could. It is apparent that whether or not the precise time pattern argued holds and holds for all SAC missions is an empirical question.

In the major multi-strike campaign analyses, not much attention has been devoted so far to the exact time development of the campaign.

Systematic work on this has been postponed perhaps because for choice among carriers, which is what is at issue in most systems studies so far, the length and pattern of time development of the campaign may not be crucial. For choice among bases it is of great importance.

Some assumptions in RAND systems studies which affect campaign length, although perhaps not weapons choice are, (1) the number of targets in the target systems, (2) the amount of target identification and BDA reconnaissance required, (3) target recuperability, (4) the levels of enemy local and area defenses (and more particularly, the ratio of local to area defenses), (5) the assumption of simultaneous attacks against much of the target complex (To the extent that the defenses of target areas in various parts of the SU are assumed to be mutually independent, attacking different target areas on successive strikes is feasible.), (6) the probability of obtaining aircraft in the numbers that might be required. (A weapons comparison study should not be constrained by current production plans since presumably the study can have an effect on production decisions, but in a study of the relevant factors in base location, we must be less sanguine. Even the proposed 138-wing program, which has not yet been accepted by Congress, provides no more than a fraction of the aircraft needed for most MACS campaigns.) (7) The interval between strikes which is not only affected by the two-thirds availability policy covered above, but also by the amount of damage returning aircraft can be expected to have suffered and presumably by the length of the mission. (With losses per strike in the neighborhood of 50 percent, the time required to bring the surviving aircraft to a combat ready status would probably take longer than the average of four days often assumed.)

The effect of changes in these and other relevant parameters has not yet been investigated quantitatively. As a result, there is no basis for quantitative judgment as to best time patterns. It is clear that the answer to questions as to optimal time patterns will vary with the type of mission.

(e) Time patterns for various SAC missions. We tend to think only about the bombing of industrial target systems when we talk about the time involved in a SAC campaign. It is a good idea to consider SAC missions more broadly. The following running comments concern reconnaissance, counter-air, retardation, industrial, and city targets.

1. Reconnaissance. Reconnaissance missions have many formal similarities to bombing strikes and the corresponding mathematical models have much in common. In the ensuing comments on time, their difference is what interests us.

   By definition, pre-strike reconnaissance precedes bombing. Moreover, as we understand it, a large proportion of SAC targets will require it. MACS intended to stage a campaign including reconnaissance for 50 percent of the targets of a known target list totaling 100. For known target lists as large as 1,700, of the sort presently talked about in the Air Force, a predominant majority of the targets would require reconnaissance. Furthermore, general exploration of the SU has seriously been considered by SAC to be necessary in order to discover targets like bomber bases and atomic weapon plants. This sort of search requirement is obviously even more extreme. If there are any over-all desirable time limits on the bombing campaign, they operate much more restrictively on reconnaissance. Conversely, in this circumstance drawing out the reconnaissance period would make the bombing period still more critical by either leaving less time for appropriately briefed missions or by forcing an increase in the number of bombing missions required through inadequate briefing.

   The amount of bomb damage assessment reconnaissance that might be required is, at present, uncertain. If gross BDA cannot be accomplished by radar observations, or by photographs taken from bombing aircraft, additional missions must be flown, or the target must be reattacked and possibly over-killed.

   This is underlined by the dependence of reconnaissance on the weather. It is generally assumed that cloud cover of less than 20 percent is required. According to the weather studies which MACS is using, this occurs from 10 to 40 percent of the time depending on the season.[9] This means that reconnaissance is subject to delay of a kind that is extremely difficult to predict. The uncertainty means that some reconnaissance missions will have to be flown many times.

   These considerations all suggest high rates of reconnaissance very early in the campaign for targets like counter-air targets and retardation targets, whose bombing needs to be early.[10] For targets later in the schedule, the accomplishment of reconnaissance is one step more urgent than the strike; and the sending out of the first reconnaissance mission in random relation to the weather, to photograph the targets is in general several steps more urgent. If special missions are flown in order to gather weather date, subsequent photo-reconnaissance missions may not be mounted in random selection to the weather, but in the schedule of urgency the weather mission must then precede the target identification mission as well as the bombing mission.

   A relevant point of difference between reconnaissance and bombing is the obvious one that reconnaissance does not involve expenditure of relatively scarce A-bombs. It does not mean "emptying the magazine." Consequently, the need to husband this resource, which Brodie stresses, does not act to prolong the reconnaissance period. Strategic reconnaissance, if we understand it correctly, will represent a considerable part of the cost of a SAC campaign. A reconnaissance mission may be as costly as a bombing mission. It takes the same number of escorts to get in and to get out, etc. The effect of night operations on the cost of reconnaissance is not clear. There are two effects operating in opposite directions. While photographs may be taken at night, the area observed is substantially less than during daylight and a premium is put on the ability to navigate to the expected location of the target. However, with observation only directly below the aircraft possible, cloud cover of greater than 20 percent may be allowable. Under any likely circumstance, reconnaissance, it seems, will be expensive, and, if a large proportion of the targets will require either pre-strike reconnaissance or BDA, then the appropriate mission pattern for reconnaissance missions is an important matter.

2. Counter-Air targets. It is clear from considerations of symmetrical air war that the strategic destruction of the enemy's air power is of great importance and the order in which his and our strikes occur against our and his air bases is vital. The Air Staff takes his long range air force as our highest priority strategic target.

   It is also clear that the recuperability of air bases, especially dispersed air bases, with a considerable passive defense, must be taken into account. The intensity and frequency of attack are relevant for such considerations. Our study suggests that recuperability is an important possibility for our own bases, and this emphasizes the importance of repeated attack over time. With the passage of time, however, we might suffer cumulative damage on the ground, and unless the Russians have obligingly matched our own slow rate of destruction, this would affect systems cost.

3. Retardation targets. Considerations closely analogous to those that pertain to counter-air targets apply here.

4. Industry targets. Here the time of destruction is not as critical for forestalling the functioning of the object attacked. The function is one of production, and there is a longer time lag in its effect on the Soviet military effort than in the case of retardation and counter-air targets. Nonetheless, time is of some importance and some industrial targets are quite analogous to counter-air and retardation targets. In particular, petroleum. If the relation of production to stock and consumption in the Russian case is anything like that obtaining in our own, petroleum is an important retardation target.

   Again, the recuperation time for an industrial target must be considered if we are to obtain a realistic picture. In most RAND analyses it is assumed that a target killed is killed dead. There is, however, the problem of resurrection. This takes longer than three days, but according to the indications of Marc Peter and others, it is not a miracle. In both Nagasaki and Hiroshima the destruction of buildings frequently left the machine tools largely intact. These of course were baby bombs, but nonetheless some caution is indicated in judging the finality with which targets are killed. Marc Peter surmises that perhaps as little as three months might be the average recuperation period for industry targets after atomic attack, and he emphasizes the necessity for considering an "open" target system with killed targets re-entering after recuperation. The recuperation time of a given target is itself a function of the frequency and size of the attacks against the total industrial target system, the number of targets destroyed in a given time.

5. City targets. These may be considered with economic effect uppermost in mind. Then similar comments apply as in the case of industrial targets. The Social Science Division, however, has correctly emphasized the primacy of political effects here. These are the primary objectives analyzed in Warbo.

In connection with Goldhamer's suggestion of maximum interval of a week one might ponder some recent Air Force studies in which the sortie rate of air-refueled medium bombers from the ZI is assumed to be two missions per month.

The numbers indicated as to the desired length of the campaign, for example, would not necessarily command the agreement of other members of the Social Science Division. However, the orders of magnitude suggest the possible importance of frequency of attack for political objectives and therefore the weight assignable to variations in sortie capabilities of various base locations.

Most importantly, Goldhamer as well as other members of the Social Science Division would very probably feel that a lot of work needs to be done on the cost and effects of various campaign time patterns before best patterns can be chosen for city targets. We agree. This is the essential point.

III. Summary and Conclusion

Second, on the importance of sortie rates in various campaign time patterns:

1. For some important SAC missions, time constraints operate critically and here locational variations in sortie rate will affect systems costs.

2. For other SAC missions, how critically time constraints operate is an open question. On this much work remains to be done at RAND, and in this work the cost effect of variations in sortie rate will play a role.

3. Given the time patterns posited by the Air Force, there is no doubt that the sortie rate is important for the entire SAC task.

[2]D-1114, "Economic and Strategic Considerations in Air Base Location," Albert Wohlstetter and H. S. Rowen. Section 7, pp 7-1, and 7-2.

[3]Section 7B on generalized bombing and combat radius explicitly presents cost per bomber as well as costs per sortie. Section 13 in comparing the ZI base complex with the "least-sortie-cost" complex fixes the quantitative contribution of the sortie rate variation as a factor in the costs per sortie shown there.

[4]See D-1114, Section 7C; Section 13, p. 4 ff.; Section 14, p. 1; also Cf. Armen Alchian's M-5124.

[5]In the example given (Limestone-Leningrad), the four-year cost of a refueling base (in Iceland) is approximately $40 million. Such a base has the capacity for landing, refueling, and permitting the take-off of 25 aircraft in about an hour. With bombers refueled by identical tankers and one tanker per refueling, the cost per bomber for the refueling capability is approximately $39 million. The ratio of costs is about 39:1.6 or 24:1. With an added refueling base for insurance, the ratio would be about 12:1.

[6]Pp. 13-3 and 13-4.

[7]Whether membership in the overseas complex is selected on a least-sortie-cost or least-cost-per-bomber basis, the weighted average cost per bomber is about the same.

[8]On the limitations of our estimates of primary base defense, see D-1114, Section 10, pp. 2, 3, & 12; Section 13, p. 5; Section 14, p. 1.

[9]"Days Favorable for Aerial Reconnaissance over USSR;" and "Climatic Factors Affecting Aerial Reconnaissance over the USSR;" prepared by the Air Weather Service.

[10]Where prior information is probably at a minimum.