Jeffrey LewisForden on Safir-e Omid

I am back from Ithaca.

Funny story: Ted Postol walks into our meeting, laughing, haven taken a phone call from a reporter. The poor man was desperate, having tried to track down Ted, Geoff Forden, David Wright, me and others — but no one is returning his calls. He can’t figure out what happened to us, because no one is in the office.

Turns out, we were just at the same technical workshop on space security, hosted by George Lewis. We were able to juggle the schedule around to talk about the Iranian space launch. Goeff Forden kindly sends along a summary of his talk:

President Ahmadinejad of Iran has a much appreciated habit of going to his country’s most advanced centers of technology and having his picture taken in front of the most interesting items there. Not only does this serve to advertize the advances Iran is making in a variety of technological fields (his visit to the Natanz enrichment facility comes to mind), but it provides a wonderful opportunity to analyze just how far Iran has developed! In the case of his visit to the Iranian Space Center last February, the images posted on his website provide convincing evidence that Iran is breaking off of the SCUD-type rocket technological arc and developing a number of important advances in rocket technology.

The most important photo shows Mr. Ahmadinejad, standing next to a piece of equipment labeled in Farsi as Second Stage. It appears to be a static test version of a two engine cluster where the engines share a common turbopump. This in itself is an important technological advance! However, by itself, it would not indicate that Iran was advancing beyond SCUD technology. Also visible below the mess of piping (which is very disorderly, another indication that this is just a development model for use on a static test stand) is what appears to be two hydraulic jacks, one associated with each of the engines. These could clearly be used for moving the associated rocket engine back and forth to control the direction of each engines thrust—something that is known as thrust vector control or TVC. (In another photo, such jacks are seen displayed on a table other components that could be associated with TVC; a display reminiscent of the centrifuge components displayed for Mr. Ahmadinejad at Natanz.) This change to a gimbaled engine TVC is the first important advance beyond SCUD technology that we can positively attribute to either Iran, North Korea, or pre-war Iraq!

SCUD missiles, and their direct descendents such as the Shahab-3 (or Nodong as its North Korean variant is known as), which the Safir uses as a first stage, use graphite jet vanes fixed to the bottom of the missile and stuck into the exit exhaust. When the guidance unit detects an unwanted tilt to the rocket, or when the pre-programmed pitch program calls for a change in direction, a signal is sent to the servos that control the angle of these jet vanes and the rocket is set back on the proper course. However, sticking these vanes into the exhaust robs the missile of about 5% of its thrust. They also limit the size of the rocket engine. The next stage of development of TVC is to mount the engine on a gimbal and actual tilt it in the appropriate direction.

These jacks, however, appear to be only for tilting the engines along a single axis and would, therefore, be difficult to control both pitch and yaw. It would be more appropriate for Iran to mount four of these rather small engines on the second stage; one pair of engines mounted opposite each other across the central turbopump could act in coordination to control pitch while the other pair controlled yaw. That, of course, would also increase the thrust of the second stage and is what I assume for modeling the capabilities of the two-stage Safir.

The engines shown in this photograph are small and clearly different from either SCUD or Volga (the engines used for both the SA-2 SAM and the Iraqi Al Samoud). The question is, would Iran develop a new engine for the same sorts of fuel the SCUD uses or would they develop a new engine that used a more powerful fuel/oxidizer combination. While this cannot be determined from the picture, it makes the most sense to me that they would develop a new engine for UDMH/IRFNA combination. This has a significantly improved specific impulse and, again, is what I use to model the performance of this rocket. The dimensions of the second stage can be determined from comparing its length to width and using a diameter of 1.25 meters (there is, of course, some disagreement about this value!) These dimensions, which yield a total length of 3.5 meters for the second stage (not including the nose fairing) are illustrated on the picture below:

Could this two stage rocket put something into orbit? My calculations, with the optimistic assumptions I’ve made, indicate no, it couldn’t. However, it can reach an altitude of roughly 200 km with a speed of 6.5 km/s (it would need a velocity of roughly 7.5 km/s at that altitude to reach orbit). That’s actually very impressive! A small solid-propellant third stage inside the nose fairing with a light satellite on top could reach orbit. However, to get there, I had to assume that the second stage coasted for about 110 seconds from first stage burnout to second stage ignition. This causes considerable problems for second stage since it would require it to ignite without the benefit of a force pushing the fuel and oxidizer into the second stage turbopump. If Iran could deal with this problem, it would be another major advance for them!

However, the second stage central turbopump provides a natural solution to this! If the solid propellant gas generator causes enough “exhaust” out the central manifold, it might be enough to push the fuels into the turbopump as that gets started. After all, SpaceX’s Falcon 1 uses the exhaust from its turbopump to control unwanted rotation. It turns out that the fact that the USS Russell saw the second stage veer wildly off course, might give us some clues as to not only how the second stage failed, but if there was a coast period built into the trajectory (and indicate if Iran was trying to ignite a liquid propellant motor in free fall.)

The Russell, based in the Gulf of Oman, can only observe the Safir with its radar after the rocket has risen a considerable distance above the Earth’s surface. In fact, if the Safir’s second stage ignited at the time of the first stage burnout, and used the acceleration of the first stage to push the fuel into the second stage turbopump, then it would be almost three degrees below the Russell’s horizon and invisible. If, on the other hand, the Iranians tried to ignite the second stage in free fall, then the Russell could have observed the attempt about 2 degrees above its horizon.

Why did the Safir second stage veer wildly off course? Two possibilities immediately suggest themselves: 1) the new TVC system failed and 2) the attitude control/determination system failed. Either of these, and in deed some other possibility, is possible. However, since we know it failed after a long coast period, I tend to favor the failure of the attitude control/determination system.

I missed this, but back in February the ISNA and Fars news agencies posted a bunch of pictures of Ahmadinejad walking around the new second stage engines for the Safir.

Here are the Fars links:

at[insert image number]

Her are the ISNA links:

Check ‘em out:


  1. EarthenBerm (History)

    So… how long until these photos are off the Iranian website, do you wonder?

  2. mike (History)

    I dont know for sure.. is he making a rabbit on the screen or making his thumb disappear?

  3. Magrie Leber (History)

    Um…Wouldn’t two hydraulic actuators imply two-axis control? They seem to be mounted 90 degrees apart.

  4. Yale Simkin (History)

    Geoff was quoted:

    “This change to a gimbaled engine TVC is the first important advance beyond SCUD technology that we can positively attribute to either Iran, North Korea, or pre-war Iraq!”

    This may be the first advance we can positively ascribe to these programs, but NK appears to have already made the advance.

    The later model TaepoDong series and Nodong-B have been widely understood to use gimbaled vernier engines to steer at least an upper stage.

    Yale Simkin

  5. Yale Simkin (History)

    My last sentence should have read:

    The Nodong-B variant and the later model TaepoDong series (at least in an upper stage) have been widely understood to use gimbaled vernier engines to steer.

    Yale Simkin

    (An after-posting editing function would sure be swell!)

  6. Geoff Forden (History)

    There are signs that Iran might launch another rocket very soon; perhaps even today. If so, I find it hard to believe that they would have had time to analyze and correct whatever went wrong with the last test. Still, stranger things have happened!

    By the way, it seems likely that the Safir is the civilian version of the Ashura missile that the Pentagon found so worrying last November. When I model the Safir with a 1000 kg warhead, I get a maximum range of about 2,600 km.

    Also, I seem to have taken for granted that the policy implications of Iran breaking out of the SCUD-technology arc would be obvious. Let me emphasize this: Iran is no longer dependent on North Korea for missile technology and in fact we could expect to see missile know-how flowing in the opposite direction. Also, SCUD technology is basically a dead end: ICBM’s based on SCUD-technology would have to be enormous! By assimilating missile technology, Iran is taking the first steps toward having a practical ICBM at some time in the future. It should, of course, be recognized that the Safir (or the Ashura) is not that missile; it will be many years before Iran has the technology to deliver a nuclear warhead to the United States. But they will certainly have that capability one day.

  7. Geoff Forden (History)

    Vernier engines are an improvement over jet vanes but they are not the sort of quantum leap that gimbaled main engines are. And virtual missiles (later model Taepodong and Nodong-B variants) are not the same as real missiles (Safir).

  8. Geoff Forden (History)

    It appears that early reports that placed the USS Russell in the Gulf of Oman were incorrect. Recent reports simply say that the Russell was in the Persian Gulf. (Apparently debris from the launch rained down in the Gulf of Oman.) Assuming the Russell was off the coast of Kuwait implies that its SPY-1D radar could see the Safir rising above the horizon within seconds of the first stage burnout; I estimate about 100 seconds after launch. That is considerably sooner than my estimate for when the Russell could see the Safir from the Gulf of Oman (which was about 140 seconds after launch) and therefore there is less reason to believe that the Safir second stage had a long coast period before it ignited.

  9. Yale Simkin (History)

    Geoff wrote:

    “…virtual missiles (later model Taepodong and Nodong-B variants) are not the same as real missiles (Safir)”

    I was not making myself clear. I was referring not to planned, but actual production rockets – the NoDong-B, and the Taepodong-2C.

    Gimbaled verniers may not be identical to gimbaled main engines, but they are a technological leap past the Nazi V-2 and its offspring SCUD and Redstone steering-vanes.

    Yale Simkin

  10. Geoff Forden (History)

    Yale’s insistence that there are real missiles in production named Taepodong-2C and Nodong-B and that we know they use vernier engines for their second stages is symptomatic of a malaise in our field. Too many people are willing to accept the wildest claims about details of missiles that have never been flown. I suspect it is caused by the proliferation of pretty drawings on the web of weapon systems that have little, if any, supporting evidence.

    Take the Taepodong-2 for instance. The North Koreans have tested one missile the West calls a Taepodong-2 and it failed 40-some seconds into its flight; long before its second stage could be ignited. How could anyone know what a missile system might look like two variants away from the one and only test flight much less what its second stage might use for thrust vector control? I have struggled to even get an estimate of the gross properties of the Taepodong-2 that was actually flown and I know there is very little evidence on which to even base an estimate on its total mass.

    The pictures the Iranians have released are the first real evidence that nations, just starting on their development of indigenous missile technology, have progressed beyond the arc of SCUD-technology. I have taken those pictures and tried to infer their direction. Quite a bit is hypothetical even if it is based on pictures. For instance, there is no evidence that the small engines that are certainly different from SCUD or Volga engines, use UDMH. I acknowledge that but point out that it makes little if any sense for them to develop a new small engine without developing it to use a new, more powerful fuel/oxidizer combination. (After all, the engines are the approximate size as the SA-2 Volga engine and it would make more sense—in terms of reducing development risk—to use an existing engine if the fuel was the same.) But my assumptions are clearly stated and people can evaluate them for what they are.

    Let us try to base our discussions, which have important policy implications, on actual facts and not flights of fancy.

    ps: I have no desire to single Yale out since I have a great deal of respect for him. I am trying to point out a wide spread trend.

  11. Yale Simkin (History)

    Geoff wrote:
    “I have no desire to single Yale out since I have a great deal of respect for him.”

    Not to worry… I agree with
    “There is only one thing in life worse than being talked about, and that is not being talked about.” Oscar Wilde

    Geoff wrote:

    “Yale’s insistence that there are real missiles in production named Taepodong-2C and Nodong-B and that we know they use vernier engines for their second stages…

    I think he is crediting my comments with much more certainty than I intended.

    Geoff described the turbo-pump and possible gimbal arrangement of the main engines. To which I commented:

    This may be the FIRST advance we can POSITIVELY ascribe to these programs, but NK APPEARS to have already made the advance. The Nodong-B variant and the later model TaepoDong series (at least in an upper stage) have been widely UNDERSTOOD to use gimbaled vernier engines to steer.

    Rather hedged with little insistence.

    A typical example (which is chock-a-block with caveats) may be found in Charles Vick’s analysis at Global Security.

    Here is an edited down portion of the relevant table:

    There is always the tendency to create conclusions which are stronger than the facts available. Analysing situations that range from translucent to opaque is a messy process.

    Yale Simkin

  12. Markus Schiller (History)

    We at Schmucker Technologie fully agree with the statement of Mr. Forden. This malaise regrettably is characteristic for the field of missile analysis while actual facts are ignored.

    Now a short commentary regarding the upper stage propellants:

    It is improbable that UDMH is used in this stage for several reasons, the most evident being the size of the oxidizer and fuel tanks. The tank size ratio of missiles using UDMH/IRFNA (such as R-16 for example) is roughly 1:1. The Safir upper stage has a ratio of about 2:1, which is quite characteristic for kerosene/IRFNA (Scud B, …). If this is the case, then I would have to correct Mr. Forden’s caption of the upper stage image: The small tank in front probably is the fuel tank.

  13. Hairs (History)

    Geoff writes: “…it makes little if any sense for them to develop a new small engine without developing it to use a new, more powerful fuel/oxidizer combination.”

    I know enough about rocketry to almost fill a medium sized post-card, so in general I’m happy to defer Geoff’s view that it makes sense for the Iranians to develop a new engine along with the gimballed mounting. But I question whether it makes sense to test both developments at the same time in the same flight. Would it not be better to first test the gimbal mounted nozzles with the old motor and fuel, and to test the new motors and fuel with the old graphite vanes?

    Having spent many years of my life commissioning leading-edge process equipment I know how hard it is to disentangle the bugs and influenecs of too many new developments all stuffed into in one product. I can easily imagine that the Iranian rocket engineers would like to do a few tests of each development on its own (coupled only with an already understood piece of equipment) before putting the two together. In which case it would perfectly plausible that the photos show “just” a gimballed version of an “old style” engine, while behind the photographer’s back sits the brand new motor and fuel design stuck in a casing with control vanes!

  14. Geoff Forden (History)

    I fully agree with Markus Schiller’s analysis of the implications of the tank sizes and the use of kerosene/IRFNA and Hairs makes a very good point about the logic of only a limited number of changes.

  15. Hairs (History)

    Geoff, Markus:

    If I may first set the scene for others as ignorant in rocketry as I am…

    A little googling shows that kerosene (ca. 750 kg/m3) and UDMH (ca. 785 kg/m3) have similar densities and, when combusted with IRFNA, have similar specific impulses (kerosene = ca. 256 s, and UDMH = ca. 266 s). The two biggest differences I can see between them are:

    1) The UDMH reaction is hypergolic (i.e. self-igniting) whereas the kerosene reaction is not, and

    2) The oxidiser to fuel ratio is about 1.3:1 for UDMH and about 4.4:1 for kerosene.

    All of this suggest that per kg of combusting material the kerosene reaction requires relatively more oxidiser (which is comparatively dense at around 1500 kg/m3) and also requires an ignition system (at least in a second stage).

    So, it would seem that the kerosene reaction requires a higher “average” density of fuel and oxidiser, and also requires an additional system (or, at least, gives up the – assumed – greater simplicity of autoignition). Therefore what puzzles me is: What is the advantage of moving to a kerosene design? Is the kerosene reaction so much more energetic that it easily offsets the extra mass of oxidiser and ignition system, or is it some other factor? Surely the Iranians wouldn’t go to all the effort of designing and testing a kerosene system just because it’s a bit nicer to handle, would they?

    From what I’ve read (admittedly limited to a rather short lunchbreak!) UDMH plus IRFNA was an early- to mid-fifties technology generally followed by a change to N2O4 oxidiser in the late fifties. So what could be the Iranians’ motive in going to kerosene, if that is what is suggested by the relative sizes of the tanks?

  16. Markus Schiller (History)


    All other Iranian liquid systems that were actually launched until now (for example Shahab 1, …) use kerosene as fuel, so they are familiar with it.

    As you pointed out, total propellant mass of a UDMH system is less than that of a kerosene system of the same size. This might neutralize the advantage of higher UDMH Isp – at least for same sized systems. When you try to neutralize this effect again by increasing your tank size, then you get a weight penalty for your rocket stage that again counters the potential UDMH advantage. But these are just basic thoughts that need further analysis for reliable statements.

    Anyway, if you fill the propellant feed line with tonka before you fill the rocket with kerosene you have a nice hypergolic combination for ignition – the Scud-type way of engine ignition.

    Combined with the nice handling of kerosene that you mentioned these additional aspects further favor kerosene against UDMH.

    So, if you have kerosene designs at home that work, you stick with them.

  17. Geoff Forden (History)

    There is a difference between “old” technology and technology a country is familiar with. The Iranians (and indeed all those countries that got started with missile production using SCUD) are familiar with “kerosene” (actually, a fuel very similar to kerosene, we are using kerosene as a kind of short hand) and are not familiar with UDMH even though more advanced missile developers moved on to UDMH many years ago. The Iraqis are a good example of this. They tried for many years to modify SCUD engines so that they could use UDMH but were always unsuccessful; the increased temperature of burning UDMH simply melting the combustion chambers (and possibly the turbopumps as well). The fact that the Iranians didn’t design a totally new engine should indicate how difficult it is to build a new engine, especially when you realized that there are dozens of associated production plants for different components that would also have to be changed.

  18. Hairs (History)

    Geoff, Markus:

    Can one usefully infer anything about the fuel / oxidiser combination by looking at the size and shape of the nozzles? For example, the mixing ratio of N2O4:kerosene (actually “RP-1” appears to be the common term for kerrosene-type fuel) seems to be much the same as IRFNA:kerosene, so IN PRINCIPLE the tank ratios could represent a move towards a different oxidiser.

    Presumably if Iran is buying foreign assistance in developing its rockets then its technological changes and improvements are likely to mirror what happened elsewhere in earlier years. For example, if Russia / China / Pakistan / India developed TVC at the same time as a move to N2O4, then the proven designs and technical know-how available to be bought (either from a state wishing to cash in on, for them, an obsolete design, or from impoverished pensioners hoping to sell their lab books and knowledge from 50 years ago) would also bundle TVC and N2O4 into the same design.

    Unless there’s evidence to the contrary I quite happy to accept that Iran is sticking with developments based on kerosene:IRFNA motors. But it would certainly be handy if it were possible to make statements like: “The angle of divergence of this nozzle is incompatible with propellant X and oxidiser Y”, and consequently put some additional constraints on what we believe the Iranians’ capabilities are.

  19. Jochen Schischka (History)

    1.) To Hairs:

    I don’t think there is a way to guess from the form/size of a nozzle which kind of fuel an engine uses; But the area ratio end/throat tells you if this is generally a ground- (around 10) or a vacuum-optimized (20-50 or more) nozzle.

    2.) To Geoff:

    Only the non-cooled nozzle-extension melted in the iraqi UDMH-experiment, neither the chamber nor the turbopump (there is a nice photo in the UNMOVIC-Compendium, Chapter IV, page 442)!

    3.) To anybody:

    Is it possible that the upper-stage engine of the Safir IRILV might be identical with the two-chamber-vernier-engine of the R-27/4K10/RSM-25/SSN-6/Serb? In my opinion, that device would fit quite nicely: approximately right size, “obsolete” (i translate this as a) “proven”, b) “robust” and c) “available”) russian technology, two one-axis-gimbaled chambers, one open-cycle turbopump (submerged in the lower tank – thus modifying the visible tank-volume ratio, which is also further distorted by the common bulkhead of both tanks), Oxidiser NTO (although i would expect a high HNO3-share – something like “AK-55”?), Fuel UDMH (maybe also mixed with some other component?); The only two grains of salt would be the relatively low thrust of about 3 tons (at ground level?) and the fact that this solution would require five different kinds of fuel (first stage: NoDong-engine with TM-185/AK-27I/TG-02 + second stage: Serb-Vernier-engine with “NTO”/UDMH). Based on photos from

    i’d also say that the Iranians seem to be planning to implement a non-cooled nozzle-extension (like the Iraqis several years earlier…maybe with analog results!), thus optimizing the engine for high-altitude use.

    With some photo-measurement, a body-diameter of 1.25 m and a little Sunday-afternoon-CAD-modeling i get a fuel volume of about 1.46m³ and approximately 1.77m³ oxidiser, thus an O/F of something like 2.2-2.3 if you follow my theory on the R-27-heritage, and a need of around 60 liters of pressure-gas at 2 MPa – as i understand, this would be consistent with a toroidal tank from the Scud-C/Shahab-2.

    Has anybody exact information on the R-27 and its specific fuels and photos of the engine for comparison? I can’t find any useful data for this missile on the internet.

    By the way, this idea could also elegantly explain the rumors about “SSN-6” in North Korea/Iran…