Find a Security Clearance Job!


Titan Costs


Titan costs have been driven up almost 60 percent -- approaching $325 million for a Titan IV Centaur. Inefficient production rates primarily account for the increase in Titan IV costs -- Titan production was originally sized to support a production rate of 10 per year versus today's rate of 3 per year.(2)

All User funded Titan IV vehicles and cost increases to Air Force funded vehicles after December 1992 are incrementally funded and production on each vehicle is not completed until vehicle is prepared to be shipped to the launch sites. Therefore recurring Flyaway dollars do not correspond logically to procurement quantities. The costs for launch processing are based on actual contract values for the current Titan IV program and were transferred from operation and support costs to procurement costs in conjunction with the FY92/93 President's Budget. Thus, these costs are not included in the range costs, which continue to be carried as operation and support costs. Range costs are based on current and historical data from the Titan IV and Titan 34D program. The Titan IV 1994 Program Office Estimate (POE) total annual O&S costs were estimated to be $52.2M in base year (FY 1985 Constant) dollars. With a reasonable rate of four launches per year the average annual cost per launch in base year dollars is $13.05 million, which is nearly double the Titan 34D average annual cost per launch in base year dollars of $7.5 million.(3)

ELV costs are widely regarded as being effectively independent of flight rates (at least at anything resembling current flight rates of current launch vehicles). This belief is substantiated by the observation that public discussions of ELV economics are framed entirely in terms of pricing (which has as much to do with negotiating stratagems as with underlying production costs), and that ELVs are characterized as having costs/prices that are stated independently of flight rate. The Air Force estimates the cost of a no-upper stage Titan 4 at approximately $177 million, and approximately $230 million with Centaur upper stage.

The Selected Acquisition Report data provide a number of data points in the 4-6/yr annual build rate, as well as a number in the 2/yr rate. The data is fairly "dirty," so there is considerable scatter, and it separates recurring and non-recurring costs in ways that are not transparent, and the annual budget numbers do not directly correspond to the annual build-rate, so the conclusions are illustrative rather than conclusive, but it does serve to bound the problem, by simply taking both the recurring and non-recurring costs as a single annual number. These calculations omit those years which include no unit production (they have Y axis data but nothing on the X axis). This tends to underestimate fixed costs, since there are long stretches in which there are substantial expenditures with no units (~$5B FY2010-14, with final 2 units procured in 2009).

The higher build rates cluster in the $150 to $250 million unit cost range, and the 2/yr range from $400-750 million. But it is possible to bound the data with curves based on assumption concerning fixed and variable costs. One set of curves that bound the data derive from assuming that there are no marginal costs and only fixed costs, which may be estimated at between $1 billion and $1.5 billion annually. Another curve that bounds the data on the low side assumes annual fixed cost of $750 million, and marginal unit costs of $50 million, while the same fixed costs with marginal costs of $100 million bounds the data on the high side. Fixed or marginal cost assumptions significantly outside these ranges fail to bound the SAR data. Based on these rough calculations one may conclude that Titan marginal production costs are in the vicinity of $50-100 million, and fixed production costs somewhere between $750 and $1.5 billion (there are also annual fixed operations & maintenance and military personnel costs of some $100-250 million).

Titan 4 was originally planned for a launch rate (and thus production rate) of approximately 10/year (although Titan is facilitized for a 20/year flight rate). At this rate, any set of bounded assumptions provides a cost/launch of roughly $100 million, and there is only slight reduction with double the rate to 20/year, as presumably will happen with EELV. This is nearly an order of magnitude reduction for Titan, from $750 million each at 2/year, to less than $75 million for 20/year.

The problem with Titan-4 currently is that the production plant is facilitized for 20/year, and staffed for 10-12/year, which is why in the out years (post-2000), when the production rate goes to 2/year, the cost/flight goes to ~$1 billion (then-year dollars) or ~$500 M (FY95$), which is the entire dynamic driving EELV. Martin organized the workforce assuming about 10/year -- which is the number of people required more or less regardless of the build rate (from 2/year to 20/year).

These estimates of fixed costs may be too high, particularly since out-year projections of total DoD total ELV space launch procurements are only about $800,000,000, for all Delta, Atlas, and Titan missions combined.(4) However, these estimates include only DOD launch vehicles, and exclude NRO costs. Since NRO will be the sole customer for the Titan 4 after around 2002, this estimate does not reflect Titan 4 costs.

Older SAR data from 1990 (using a set of cost/flight data at different flight rates) suggests an expected variable cost of about $100,000,000 for the core plus SRMUs, and about $50-60,000,000 for the Centaur. With a fixed cost base of around $ 300-350,000,000, this produced an annual cost of about $800,000,000 (4 vehicles), which matched the out-year estimates of that time.

To get good numbers, it is necessary to dig another level deeper. The SARs reflect each release's current total budget projection, and as the assumptions change about the overall timing and content of the program, fixed and variable costs can be lost in the noise. The problem is the SARs are heavily weighted towards the non-recurring part of the program, and the fixed vs variable calculations really need to be done on an annual recurring basis, which is poorly reflected in a purely-SAR based calculation.

The recurring cost base (which is in the SAR) includes a set of non-recurring costs including the Titan-IV "Bridge contract" which pays LockMart for not producing Titans, and includes a variety of CPPI (Continuous Products and Process Improvements) such as testing a ablative first stage nozzle extension, new turbopumps, new engine injectors, composite engine frames, and a variety of reliability analyses. Assumedly, these are not recurring cost items, and need to be subtracted out of the fixed cost basis to get a proper out-year fixed cost estimate.

For Titan the problem is complicated as there has historically been a high level of mission and vehicle customization for particular users. This has increased the fixed cost base by adding in "sustaining support" or "sustaining engineering" as each vehicle and mission is re-analyzed, the hardware and mission profile modified to the customer's requirements. The annual cost of this type of activity may exceed $500,000,000.

The primary customer (NRO) for the Titan-IV class mission has been spending rather freely on "customizing" the vehicle to specific payload-unique requirements. This should not be considered part of the fixed costs (as it should go away after a specific mission). But a SAR-based calculation will lump this, plus the non-recurring cost into the fixed cost base, and over estimate the fixed cost component. But presumably these are part of the fixed cost, in the sense that they are part of the cost of keeping T-4 in existence (though there are similar issues involved in computing STS fixed costs.

Another indication of fixed/variable is the transfer of Atlas to Denver. Something like 1,700 jobs moved within the first six months (overhead) and something like 400 stayed in San Diego to keep producing boosters (variable cost of actual production). According to the press reporting the 1,700 who moved were not direct production (which is why they could move so quickly), and that the 400 who stayed were production. While a breakout of jobs moved/stayed may not be a good indicator of fixed/variable costs, there do not seem to be any "good" indicators, period. But this is one of the few data-points available.

Most of the workforce is in "overhead" type coordinating positions which are a fixed cost which is not volume-dependent. These include all the recurring "subcontract management" jobs, which schedule and coordinate the subcontractor base, which obviously moved to Denver. And presumably the fixed/variable costs at the subcontractor tier would look more or less like those at the prime, and would be reflected in the SAR data.

However, it is unclear the extent to which a breakout of jobs moved/stayed are a good indicator of fixed/variable cost. This approach could produce an overestimate of the fixed costs -- since it does not include any fixed annual cost at subcontractors, nor recognizes any functions not tied to massive pieces of tooling (such as procurement, test, and avionics construction). It should be noted much of the GD Atlas production activities also moved to Denver. And a large component of this was never in San Diego (the engines, for example). What remained in GD San Diego was the primary "balloon tank" welding activities (which utilize some rather large and expensive tooling setups, which would be very expensive to move). Avionics and shrouds, and most other manufacturing activities are now in Denver. Not to mention all the recurring "subcontract management" jobs, which schedule and coordinate the subcontractor base, which obviously moved to Denver.

There were a few hundred jobs lost in San Diego, but for the most part all moved to Denver (there may be ambiguity between jobs moving but people staying) -- the jobs moved, but not the people. LockMart expected a substantial percentage more to move from San Diego to Denver. One of the problems they encountered was a bit of arrogance that "those California folks want to move to Denver". But the California folks said "why should I pack up and move where it snows? for a new boss?"

GD had the problem, as many of the aerospace firms have, in a aging workforce (average age in the mid-50's or so). They are very firmly rooted in a community, and its hard to uproot a family (with two career families a particular problem to move) and for people having sell houses. This produces a lot of resistance to moving.

This was complicated as GD had really pumped up the price of GD stock to about $150-200/share in the year or so before the sale of the Launch Systems Division, not including a one-time dividend payout of about $50/share. If someone had been working at GD for a couple of decades they probably had a good deal of company stock through the company saving plan. On that basis -- if someone is within 5 years of retirement, layoff pays 6 months to 1 year of severance (dependent upon time with the company), and they can continue benefits and meet expenses for a few years with the GD stock payout... why move? or at least, why jump to move? Most of the folks who moved were the younger members of the workforce, and while bright and moderately experienced, didn't have the decades of experience with the Atlas and Centaur production.


1. Adapted from: (David Anderman), (Ralph Buttigieg), (Josh Hopkins), (Jim Kingdon), (Larrison), (Pat), (John Pike), and (Allen Thomson), "RFC: EELV = Titan," thread on, February-March 1995.

2. Department of Defense, Space Launch Modernization Plan - Executive Summary, 5 May 1994, page 7.

3. Adapted from: Space and Missile Systems Center, "Titan IV Selected Acquisition Report," (RCS:DD-COMP(Q&A)823), 31 December 1993.

4. DOD, "Industrial Assessment for Space Launch Vehicles," January 1995 (which references OSD/API, "FY 96 Budget Estimate Submission" for the specific data).

Join the mailing list