You are here

June 2005 - December 2005


Posted December 19, 2005
Here is the apparent cause of structural damage (further analysis may change the conclusion):

Due to high winds, we placed the countdown on hold and began draining the fuel tank. As we drained fuel from the 1st stage tank, a faulty pressurization valve caused a vacuum condition in the tank. This caused a fuel tank barrel section to deform and suck inward. It is important to note that the root cause is an electrical fault with a valve, not structural design.

At this point, it appears that no other damage was sustained to the vehicle or the satellite. The rocket will be lowered down this afternoon and placed in its hangar for further inspection.

--- Elon ---






Posted December 19, 2005
Launch is scrubbed until early next year, as there is a structural issue with the 1st stage fuel tank that will require repair. I will provide further comment as soon as this has been carefully analyzed.
Consistent with our policy, we must be 100% green for launch with no outstanding concerns whatsoever. It is not just a matter of repairing the damage, but also understanding at a fundamental level how to ensure it never happens again. We will also do another full review of all the vehicle systems, including propulsion, structures, avionics, software and ground support systems. Therefore, I expect that the earliest that launch would occur is late January. Third time's the charm.

--- Elon ---






Posted December 18, 2005
All systems have passed their prelaunch checkout and we are go for launch tomorrow at 11 a.m. California time (7 p.m. GMT). Over the past month, we have also improved and upgraded the countdown sequence in several ways:

  • More computer controlled operations vs manual
  • Improved ground support equipment to load propellant/pressurant faster
  • Worked with range safety to speed up checkout of the thrust termination system
  • Changed to simultaneous load of LOX and fuel on both stages

The Vandenberg hold down firing countdown was 5 hours, the first Kwaj countdown was 4 hours and now we are at 3 hours. Having a responsive launch capability is important to DARPA and the Air Force (and us for cost reasons), so we've put a lot of effort into streamlining the countdown.

A C-17 buzzes Falcon for good luck







Posted December 15, 2005
The SpaceX launch date is scheduled for Monday, December 19 at 11 a.m. California time (7 p.m. GMT). Press release



Posted December 7, 2005
The new launch date is approximately December 20, depending on when the Missile Defense Agency testing is complete. As soon as we have a firm time, it will be posted on the SpaceX website.

Liquid Oxygen

Regarding liquid oxygen (LOX) supplies, we expect to have enough on hand this time to fill the rocket four or five times over. This should account for almost any issue with a particular storage tank as well as an extended hold on the pad. There is an engineering term known as a s*load. I have asked that we have at least two s*loads on hand in case one s*load is not enough.

We chartered a C-17 to fly two of our empty high quality LOX containers to Hawaii, sourced another high quality LOX container on Hawaii and put all three on the barge to Kwajalein. In addition, our LOX plant on Kwajalein has been repaired and is producing LOX on island again.

Some might be wondering why we were so dumb as to run out of LOX on a remote tropical island on the last launch attempt. Believe me, we tried hard to avoid it, but several issues conspired to create the problem:

  • The additional month of Merlin testing resulted in additional LOX boil-off on island. Even though it is stored in vacuum jacketed containers, LOX at -300F degrees does not like being on a tropical island at 85F.
  • The SpaceX LOX plant on island broke down a few weeks prior to launch, which meant we could not top up.
  • We ordered replacement LOX from Hawaii, but the container quality was poor, so only 20% of what we ordered actually arrived.
  • Ground winds were unusually high on launch day, which amplifies the boil-off rate significantly, since the Falcon's first stage LOX tank is uninsulated.
  • All of the above would not have mattered if our final storage tank did not have a small, manual vent valve incorrectly in the open position. Somewhat agonizingly, we were only a few percent away from being full. We just needed a little sip from the last tank.
  • After a while, we were able to close the vent and fill the vehicle's LOX tanks. However, we use LOX to chill our onboard helium and the absence of ground LOX to do so resulted in the helium heating up and venting back to storage. In the end, we did not have enough LOX to stay filled on the rocket and chill & pressurize the helium.

Engine Computer

The engine computer reboot anomaly was definitively traced to a ground power problem. Importantly, this would have had no effect on flight, since we switch to vehicle power before the autosequence begins. The reason it cropped up at Kwajalein was that the higher load on the longer umbilical (three times longer than in prior tests) coupled with high temperatures in Kwajalein resulted in increased resistance in the ground umbilical. This was just enough to lower the voltage below minimums and cause an engine computer reset when drawing maximum power. The same max power test was repeated on internal vehicle batteries with no problem at all.

This problem has been solved by slightly increasing voltage on the ground umbilical.







Posted November 26, 2005
The launch is scrubbed for today. As I warned, the likelihood of an all new rocket launching from an all new launch pad on its first attempt is low.

What happened was that an auxiliary liquid oxygen (LOX) fill tank had a manual vent valve incorrectly set to vent. The time it took to correct the problem resulted in significant LOX boiloff and loss of helium, and it was the latter that caused the launch abort. LOX is used to chill the helium bottles, so we lose helium if there is no LOX to cool the bottles.

Although we were eventually able to refill the vehicle LOX tanks, the rate at which we could add helium was slower than the rate at which LOX was boiling away. There was no way to close the gap, so the launch had to be called off. In addition, we experienced an anomaly with the main engine computer that requires further investigation and was arguably reason in and of itself to postpone launch.

We anticipate a new launch attempt in mid-December, depending on the timing of LOX resupply from Hawaii (our LOX plant on Omelek can only produce about one ton per day).






Posted November 26, 2005
A LOX fill problem has been corrected. Expected launch time is now approx 5 p.m. California time.






Posted November 26, 2005
Falcon 1 Launch Now Scheduled for 2 p.m. CA time (10 p.m. GMT) Today.






Posted November 24, 2005
Falcon 1 Launch Delayed by Army Range to Saturday, November 26

In order to facilitate preparations for a missile defense launch, the Army Space and Missile Defense Command (SMDC) has bumped the SpaceX Falcon 1 maiden flight from its officially scheduled launch date of Friday, November 25 at 1 p.m. PST (9 p.m. GMT). The new launch time is Saturday, November 26 at 1 p.m. PST (9 p.m. GMT).

All prelaunch preparation by SpaceX is complete. The final task of test firing the torch igniters worked on the first attempt, making us green for launch. Since our launch was bumped by a day to Saturday (California time), we will make tomorrow a shorter day and do some of the optional checkouts of systems that might cause a halt in the countdown.






Posted November 18, 2005
Falcon 1 Launch Date Scheduled for November 25

On Friday, November 25 at 1 p.m. PST (9 p.m. GMT), the Falcon 1 countdown to launch is expected to reach T-Zero. At that point, the hold-down clamps will release and the Falcon 1 rocket will begin its journey to orbit, accelerating to 17,000 mph (twenty-five times the speed of sound) in less than ten minutes.

The customer for this mission is DARPA and the Air Force and the payload will be FalconSat-2, part of the Air Force Academy’s satellite program that will measure space plasma phenomena, which can adversely affect space-based communications, including GPS and other civil and military communications.

More launch information

For video of the May hold down firing at Vandenberg, click here.

Falcon 1 Ready for Launch on Omelek Island






June 2005 through September 2005 Update

(Note: Subscribers to the email list receive the update earlier than it is posted on the website. Email address privacy is always respected.)

The History of Falcon 9

About eighteen months ago, a customer approached SpaceX with launch mass and fairing volume needs that exceeded the Falcon 5. We iterated on several different solutions, including upgrading the Merlin engine thrust and adding liquid or solid strap on boosters. All the options held significant drawbacks in cost, schedule or reliability, except one – a nine engine first stage.

By adding an additional four engines on the base and stretching the tanks, we were able to achieve a payload of approximately ten imperial tons to low Earth orbit, which is slightly greater than that of the Boeing Delta IV Medium. Going further and adding two first stages as liquid strap on boosters, like Delta IV Heavy, allowed us to place about 25 tons into LEO – more than any launch vehicle in use today.

This is very significant as it allows SpaceX to lift the full range of commercial and military satellites, as well as service the Space Station with considerably more cargo. It also maximally leverages our investment in avionics, guidance & control, structural design, launch infrastructure and the Merlin engine. Our strategy of using the Merlin engine throughout the Falcon product line is similar to Southwest’s strategy of using only 737s throughout its fleet. However, in our case we get economies of scale in both manufacturing and servicing of the engine.

Some people may wonder why we use exactly the same stage structures on both F5 and F9. From a structural optimization standpoint, it is obviously more efficient to reduce the tank size for Falcon 5. However, that would require considerable additional investment in a different set of tooling, transport and ground support equipment, umbilical tower, aerodynamic calculations, test fixtures, etc. By making use of exactly the same stage structures for F5 and F9, our investment and launch costs are minimized. As those familiar with space transportation know, reliability & cost are the real problem, not squeezing out the last bit of performance.

The Launch Manifest is Growing

In addition to the six Falcon 1 launches, we now have two customers on the Falcon 9 launch manifest, one US government customer in 2007 and then Bigelow Aerospace in 2008. With Falcon 9’s ability to place any size of satellite into geosynchronous orbit, we are seeing considerable interest from the commercial satellite sector. I’m confident that after the launch of Falcon 1, we will be able to add a number of new customers for Falcon 9.

Customer Launch Date Vehicle Departure Point
US Defense Dept (DARPA) Q4 2005 Falcon 1 Kwajalein
US Defense Dept (OSD/NRL) Q4 2005 Falcon 1 Vandenberg
Malaysia (ATSB) Q2 2006 Falcon 1 Kwajalein
US Government Q2 2007 Falcon 9 Kwajalein
Bigelow Aerospace Q1 2008 Falcon 9 Kwajalein
US Commercial Q2 2008 Falcon 1 Vandenberg
MDA Corp. Q3 2008 Falcon 1 Vandenberg
Swedish Space Corp. Q4 2008 Falcon 1 Vandenberg
US Air Force $100 million contract thru 2010 Falcon 1 TBD


Falcon 1 is Really Important

I want to emphasize that although SpaceX development is now primarily on the Falcon 5/9, Falcon 1 is and will always remain a very important part of our business. All of us at SpaceX really believe in the small satellite market and we will never turn away from it or relegate it to a back corner. I think that once the satellite market has time to adapt to its existence, Falcon 1 may very well see the highest launch rate per year of any rocket in the world.

We have also changed our pricing policy to reflect the all inclusive price of launch to make things really clear. Some people were under the impression that range and 3 rd party insurance costs were millions of dollars. Everything is now included, unless you have a really complex spacecraft or require an outside mission assurance process, and it is the same price we’ve had since 2002 - $5.9M for the vehicle plus $0.8M for the launch range, 3 rd party insurance and payload integration.

The Island of Dr. Yes

In June, after Titan IV’s launch from Vandenberg was delayed yet again to sometime in late 2005, we decided to switch the maiden launch of Falcon 1 to an island in the Kwajalein Atoll that we are leasing from the US Army. For those unaware, SpaceX has a launch restriction specifying that we cannot fly from our Vandenberg Air Force Base launch site until the multi-billion dollar Titan IV mission departs. In theory, there is a tiny chance that our rocket could go off course and damage the T-IV, which is sitting on its pad, so our ability to launch from there has been put on pause. .

The Kwajalein Atoll is essentially a huge reef that occasionally extends above water, forming a chain of islands. The biggest island is also called Kwajalein and contains almost all of the US personnel in the area. Politically, the Kwajalein Atoll is part of the Republic of the Marshall Islands, but is leased by the US Army for use as a missile test range and communications & tracking facility. As a result of hundreds of millions of dollars of investment by the Defense Department over the past several decades, Kwajalein is home to some of the world’s most powerful radar tracking and space communications systems.

Southern part of the Kwajalein Atoll with the main island of Kwajalein in the foreground

Our island in the Atoll is named Omelek and it is about halfway up the island chain on the eastern side. The Atoll’s location is advantageous for a number of reasons. Most significant is its location at 9˚N latitude, placing it much closer to the equator than the 28˚N latitude of Cape Canaveral, where our other eastward trajectory launch pad is located. That allows us to take more advantage of the Earth’s rotation and deliver increased payload to orbit. It also means that a much smaller plane change maneuver (to 0˚) is needed for geosynchronous satellites.

Kwajalein activity had been percolating along for about eighteen months, mostly dealing with regulatory matters, but it became our number one priority in June when we shifted first launch from Vandenberg to Omelek. From having only partially complete concrete foundations in June, the team has kicked butt and we now have the following in place:

  • Launch stand and vehicle erector
  • Vehicle hangar
  • Umbilical tower
  • Helium pressurization system
  • Nitrogen purge system
  • Liquid oxygen storage tanks
  • RP-1 kerosene tanks
  • A liquid oxygen generating plant
  • Dual redundant heavy duty generators
  • Office building (broken into pieces, brought over from another island and reassembled)
  • Fiber optic communications from Omelek to Kwajalein
  • Remote camera systems
  • Remote control of the launch site
  • 10k class clean room for satellite integration

The Army and Kwajalein Range Services have really stepped up to the plate to help get all this done so rapidly and deserve a lot of thanks. In addition to all the physical work that has taken place, a lot of effort has gone into ensuring that the rocket is safe for flight with a fully qualified, independent and redundant thrust termination system. It has also been great working with the Air Force and DARPA as the primary customers, together with NASA in a supporting role, for this first flight.

Launch Control and Amenities

Our launch control center is located on Kwajalein Island along with guest offices for our customers. The main island also has hotels, shops, a cafeteria and sports facilities. For potential customers out there, I should mention that Kwajalein has some of the world’s best scuba diving and snorkeling! It is literally a tropical paradise.

Omelek pier and landing ramp



Nine Lives

One of the most important questions regarding the Falcon 9 first stage is whether having all those engines helps or hurts. The key question in my view is whether or not true redundancy is achieved. If an engine fails, what are the odds that it will fail in a benign manner? If there is an engine fire or a chamber comes apart, will it be limited to that one engine or will it cause a neighboring engine to fail?

To answer the first question, we can look at US launch failures over the past few decades, for which we have pretty good data. Drawing from the Futron Design Reliability report that looked at failures from 1984 through 2004, there were a total of six failures that were due to liquid rocket engines. Of the six, only one failure was caused by a rupture in the combustion chamber. The other five were either feedline problems or a failure to achieve or maintain full thrust.

Thrust and blocked/frozen feedline issues are no problem for a nine engine vehicle. All it would see is a slight decrease in total thrust, which might result in a slightly lower orbit than desired (if we were at maximum payload). This is definitely a significant advantage compared with a single engine vehicle that would almost certainly be out of luck.

Then there is the question of dealing with the comparatively rare case of a chamber rupture. To protect against this, Falcon 9 will have a blast shield protecting the entire base of the vehicle just above the gimbal joints of the engines. In addition, there will be fireproofed Kevlar fragment containment around each engine, similar to those present in jet engine nacelles. The explosive power of a liquid rocket chamber is actually not exceptionally high – it can be thought of as simply a small pressure vessel containing (in our case) 800 psi hot gas. During the development of Merlin, we saw several of what we refer to as RUD (rapid unscheduled disassembly) events and no fragments have ever penetrated more than 2mm of aluminum. Also, the direction of fragments is in a shallow downward cone away from the vehicle.

As additional measures of protection, all propellant and pneumatic lines have either pre-valves or check valves nested up high in the thrust structure. If anything happens to the engine, the flight computer is able to cut off all propellant and pressurant flow immediately.

Given all of the above, I really believe we have a stage that has considerably higher propulsion reliability than a single engine vehicle.

Moreover, there are examples of multi-thrust chamber vehicles that have outstanding reliability. The Soyuz rocket, which has the longest flight history of any launch vehicle ever and a phenomenal safety record over the past few decades, is the primary form of human transportation to the Space Station. It has thirty-two thrust chambers on the first stage. The Saturn I, which had zero failures, was used for human transportation during the Apollo program. It had eight thrust chambers.

Soyuz with 32 thrust chambers

Saturn 1 with eight thrust chambers

For the stage hold down firing of Falcon 9, we will be using our very large test stand (BFTS) to deal with the roughly 350 metric tons of vertical thrust generated by the nine Merlin 1B engines. We are designing the thrust frame to be able to run to tanks dry with margin, so about 340 tons of net vertical thrust has to be held down. This is no problem for the stand itself, which was designed to handle 1500 tons of thrust! It is quite epic in stature, being about 100 ft tall with ten ft diameter steel reinforced solid concrete legs that extend 70 ft underground.

The elevator is now installed along the one leg and we have the propellant supply tanks in place. Electrical power is in place and fiber communications to the blockhouse will be installed shortly. The next segment of work is digging out the flame diverter at the base, pouring the high temperature concrete and adding the water deluge system. We are hoping to have it ready for a Q2 2006 hold down firing of the F9 flight stage.

Elevator now installed on BFTS


One of the most striking differences between Falcon 1 and Falcon 9 is that, due to the size, almost everything needs a holding fixture and a crane to be moved. Even getting in and out of the tank requires a special drawbridge structure when the manway is at the center of a 12 ft (3.6 m) diameter dome! In contrast, with Falcon 1 at 5.5 ft (1.7 m), two people can easily carry around a dome or barrel section, getting in and out of the tank requires no tooling and transportation of the whole stage is easy, as it fits inside a standard semi trailer.

At this point, we have most of the machinery and tooling built for Falcon 9. All of the domes for the first tank are in house and we are welding up the barrel sections. One significant remaining tool, due to be done soon, is the circumferential stir welding fixture. Once that is complete, Falcon 5/9 will be the only launch vehicle in the world that has fully stir welded tanks. The layup and trim tools for the composite thrust skirt and interstage are done and we are putting together our first 12 ft diameter composite barrel section.

It is astounding to me that so many rockets out there use completely different stage structures for the 1st and 2nd stages. It is a huge cost savings to be able to build the 2nd stage as simply a short version of the 1st stage. All that changes is stringer density and skin thickness.


Engine electronics for the Merlin 1B in Falcon 9 have been simplified down to just three boxes that are responsible for all digital and analog activity. Each set of engine electronics is essentially a self contained plug and play module, dealing with its own activity in accordance with high level commands issued by the flight computer on the upper stage. The only wires between the stage and each engine are an Ethernet cable and a power cable.

The flight computer issues the same steering commands to all engines, except for a slight bias on the outer engines for roll control. Unlike the Falcon 1, where the turbopump exhaust is gimbaled for roll control, the M1B engines on Falcon 9 have a fixed turbine exhaust.

M1B engine showing electronics boxes on the grey panel

The objective with the Falcon 9 avionics is triple redundancy with voting for the flight computer and inertial/GPS navigation system, and dual redundancy for the power system and telemetry, where voting isn't meaningful. Unlike Falcon 1, Falcon 9 is intended for manned flight one day and all critical systems have to function perfectly for potentially several days of occupied time.

Layout of avionics above the 2 nd stage LOX dome

----- Elon ----