A Matter of Yaw

The learning curve and early space exploration
- Part 1 -

by: Steven Kovachevich
© January 2015

The learning curve is a graphic representation of learning expressed as a function of experience. Normally learning increases with experience gained over a series of trials, repetitive behavior that increase proficiency. The learning curve can be erratic or smooth depending on the particular subject, task or number of trials. In some cases the learning curve can be expressed as a mathematical function.

The learning curve is a useful concept when applied to well known situations where past experience provides a guide. For example, the steps necessary to recover from a stall in an aircraft that are well understood. Pilots are trained to understand conditions that create a stall, recognize an approaching stall to take corrective action, or recover should a stall occur. Demonstrating competency in controlling an aircraft in a stall is part of pilot training and certification. Proficiency is gained though practice of intentional stalls that allow the pilot to gain experience with the condition.

Dealing with unknown situations is different. In these cases experience is replaced by trial and errors that build experience though incremental steps run in a repetitive fashion over and over until a particular task is understood and mastered. Potential variables encountered, understood and factored into the equation. Learning is not necessarily tied to the number of trials and a single trial can provide sufficient experience depending on the condition being analyzed.

Simply stated how well learning occurs is a function of the quality of the learning experience. The better the model simulates the actual task the better the task is performed. Where there is a large number of variables, or training less accurate to the task the greater the likelihood of unforeseen circumstances or trouble arising in performance of the task. The learning curve is a useful concept in analyzing difficulties encountered in early manned spaceflight. The large number of unknowns, lack of experience, limited quantity and rudimentary quality of training increased the likelihood of the unforseen event.

The early years of space exploration were fraught with mishaps. Missile development of the Polaris, Thor, Atlas and other rockets had multiple failures. Each failure led to another lesson learned and improved performance the next time around. These rocket failures were seen as a necessary element in the learning curve to achieve a measure of reliability and performance gained through multiple trials and errors. A similar pattern was observed in the manned spaceflight program most notably in the case of Gus Grissom's Liberty Bell 7 Spacecraft, which was lost on recovery, and Scott Carpenter's overshoot on reentry in his Aurora 7 Spacecraft. These incidents have become lore in the history of Spaceflight with the focus on astronaut performance. An analysis of the learning curve provides insight into the real causes of these two historic events.

In Carpenter's case the overshoot on reentry was attributed to a misalignment of the Capsule at the time of retrofire. Prior to retrofire Carpenter noticed a discrepancy between instrument readings of capsule position and visual ques. In the time leading up to retrofire Carpenter switched through different flight modes that confirmed malfunction of the Automatic Stabilization Control System (ASCS), a device that maintained capsule attitude while in orbit. At retrofire minus five minutes Carpenter decided to align the Capsule himself using the capsule window and periscope for attitude reference. The consequence of these last minute control activities was initiation of the retrofire sequence with a capsule misalignment of 25 degrees in yaw. The misalignment combined with a delay in the initiation of retrofire resulted in the capsule landing 250 miles downrange of the planned splashdown point. Needless to say reasons for the misalignment and overshoot immediately accrued to Carpenter and over the years have become the focus of recriminations.¹ Another look at the astronaut training schedule, evolution in simulation methods and capsule navigation techniques reveals the factors that contributed to the outcome.

Training in Project Mercury commenced almost immediately after the seven astronauts were chosen early in 1959.² The training program was based on standard aviation flight training techniques, but using simulators as actual training in space was not feasible.³ Among the various activities planned in these early months were familiarization with spacecraft configuration and escape methods, support and restraint, and operational concepts and procedures.⁴ In February 1960, the training program included two days of instruction in star recognition and celestial navigation at the Moorehead Planetarium in Chapel Hill, North Carolina. The stated purpose of the training "to assist the astronaut in correcting spacecraft yaw drifts".⁵ In May McDonnell Aircraft, manufacturer of the Mercury Capsule, delivered two procedures trainers for use in astronaut training in use of the spacecraft systems.⁶ By November 1960, the Space Task Group, in charge of Project Mercury, requested McDonnell submit a proposal for a test to determine the capability of an astronaut to make celestial observations through the Mercury Spacecraft window.⁷

A review of the Astronaut Training Summary from NASA Publication SP-45 indicates the various tasks, trainers and times astronauts spent in learning or familiarizing themselves with various aspects of spaceflight.8 The tasks were broken down into categories that included: Essential, Desirable and Questionable value. Total astronaut hours in training over the approximate two year training period was 716 hours. Of this total about 32% was dedicated to essential tasks the remaining 68% almost entirely to desirable tasks, 95% of which was airplane flights to maintain general performance. Planetarium training for star recognition deemed essential received only 28 hours, or nearly 4% of the total. The Ground Recognition and Yaw Recognition Trainers were not available until April of 1961, and used only a total of four hours, two hours each, a fractional percent of total training time. The Yaw Trainer was used by Wally Schirra for Mercury-Atlas No. 8 (MA-8) and Gordon Cooper on Mercury-Atlas No. 9 (MA-9) the last Mercury flight, but was not available for the earlier flights of John Glenn on MA-6 or Scott Carpenter on MA-7.9 The ALFA, or Air-lubricated free-attitude trainer, used primarily for attitude control training, also served the purpose of navigation training where the astronaut could maneuver through all three axis roll, pitch and yaw while observing the earth and clouds through a simulated periscope or capsule window. However, total ALFA time was only 12 hours, or under 2% of the total.

Source: NASA Flight Controllers Procedures Manual

In addition, simulation methods were crude and in their infancy. No single simulator could reproduce more than two of the conditions at any given facility.¹⁰ Effects for the ALFA Trainer consisted of the astronaut observing an illuminated strip that represented the horizon and small bulbs that simulated stars.¹¹ The Yaw Trainer consisted of a box the astronaut placed over his head with an opening that simulated the Mercury Capsule window. The astronaut would stand on a platform peering through the box opening two feet away from a thirty-three foot diameter screen on which a simulated, moving earth and clouds were projected.¹² The lack of more training time for navigation skills and late arrival of simulators to better enable familiarization with the conditions to be encountered is puzzling given one of the stated major tasks of Project Mercury was: "navigation, being able to determine the spacecraft's position in orbit at any time and determining the critical retrofire time."¹³

CEDAR 104 Project Mercury Familiarization Manuel, McDonnell Aircraft Corp, 1 Nov 1961, p.12-8

The importance of navigation in general and yaw recognition specifically was raised after John Glenn's MA-6 first orbital flight. Glenn commented on the clearly visible horizon which was vividly marked and provided a reference for pitch and roll which were easily controlled through the window. Yaw reference, on the other hand, "is not so good."¹⁴ Glenn recognized that there was a "learning period" in his ability to determine yaw. Factors necessary to yaw recognition included observation of the speed of the spacecraft over the earth which produced a drift of the ground below. The spacecraft was properly oriented when the ground moved parallel to the spacecraft flight path without drift.¹⁵ Glenn preferred the window to the periscope when determining yaw in day and night. The periscope took longer in daylight than the window and was not effective at night.¹⁶ In daylight or moonlight conditions Glenn would pitch the spacecraft to -60 degrees from normal attitude when performing this task to get a clear view of the earth and clouds below. At night yaw recognition was made with reference to the position of known stars through the window. Normal orbital and retro attitude is -34 degrees pitch, zero (0) degrees in yaw. Glenn made three 180 degree yaw maneuvers; the first a precise maneuver with errors kept to a minimum.¹⁷ The remaining two yaw maneuvers were for photography purposes only.¹⁸ The Postlaunch Memorandum Report for MA-6, dated March 5, 1962, concludes, among other things, the need to "[i]mprove the simulation of external reference systems, particularly in yaw."¹⁹

CEDAR 104 Project Mercury Familiarization Manuel, McDonnell Aircraft Corp, 1 Nov 1961, p.12-2

Scott Carpenter reported a similar experience with yaw on his flight. "Yaw reference was a problem."²⁰ Carpenter reported yaw attitude difficult to determine at night and the periscope of little use on the night side. The best yaw reference through the window was obtained by pitching down -50 degrees to -70 degrees, similar to Glenn.²¹ Carpenter performed two yaw maneuvers but only for photography purposes without a precise maneuver as performed on Glenn's Flight.²² On retrofire Carpenter experienced a 25 degree error in yaw which contributed to an over shoot on landing of some 250 miles. The experience of Glenn and Carpenter contributed to the MA-8 Mission Plan which included a series of yaw maneuvers.²³

CEDAR 104 Project Mercury Familiarization Manuel, McDonnell Aircraft Corp, 1 Dec 1962, p.13-6

The planned yaw maneuvers were designed to assess use of the window and periscope as independent references for determining attitude.²⁴ Schirra was quite successful in correcting for yaw orientation with an accuracy of about 4% error in a variety of conditions using the window and periscope.²⁵ Schirra found the periscope ineffective at night and a redundant system, which was removed for the subsequent MA-9 flight.²⁶ Unlike Glenn and Carpenter, Schirra reported yaw realignment could be accomplished in the -34 degree retroattitude.²⁷ As expected yaw determination was more difficult on the nightside owning to the limited field of view that hampered star field identification.²⁸

A view of the reticle in Wally Schirra's Sigma 7 Spacecraft.

There was no planned investigation of yaw maneuvers on MA-9 as data from earlier missions was deemed sufficient.29 The periscope was removed. In its place a navigation reticle was installed at the rear and to the left of the astronaut's window.30 The reticle was equipped with lighted lines that facilitated alignment of the capsule with reference to the earth's horizon as viewed through the window. The reticle could be stowed to provide an unobstructed view through the window when not in use. The reticle is visible in Sigma 7 on display at the Astronaut Hall of Fame in Florida. The reticle is not specifically mentioned in references to the yaw maneuvers on MA-8, but its presence in the Capsule suggests it was employed during the flight. Earlier Mercury Capsules did not carry the reticle as Aurora 7, on display at the Museum of Science and Industry in Chicago, shows. Carpenter's capsule carried the periscope instead. The reticle, if used, also appears to have helped Cooper make yaw maneuvers in the range of -20 to -25 degree pitch, a shallower angle than previously achieved, when only a small portion of the earth's horizon was visible.31

The Project Mercury Familiarization Manual, SEDR-104, a document published by McDonnell Aircraft provided a description of various capsule systems. SEDR-104 dated November 1, 1961, does not list the reticle as a navigation aid. The Familiarization Manual dated December 1, 1962, for Capsule 20, Gordon Cooper's MA-9 spacecraft shows the reticle. The MA-6, MA-7 and MA-8 Flights fall between these dates and lack reference to the reticle. Instead reference to the window and periscope predominate. The Mercury Capsule No. 16 Configuration Specification also published by McDonnell Aircraft, dated 18 January 1962, does not list the reticle as a navigation aid included in the MA-8, Sigma 7, Capsule. However, it seems unlikely the reticle was installed after the Sigma 7 Flight. Rather, the presence of the reticle in Sigma 7 suggests the device was likely installed before the Mission. The reticle a clear aid to navigation was a likely factor in improved capsule orientation experienced on Schirra's Flight.

An empty space where the reticle should be in Scott Carpenter's Aurora 7.

The conclusions of John Glenn on MA-6 that yaw determination was not so good and additional simulation techniques necessary did not impact the training schedule for Scott Carpenter's MA-7 flight. Before his flight, Carpenter spent a total of 70 hours and 40 minutes on the ALFA and Procedures Trainer; total ALFA time was only 3 hours and 45 minutes.³² Glenn by comparison reported 59 hours 45 minutes on the Cape Procedures Trainer prior to his flight covering a variety of subjects, with no ALFA time indicated.³³ Carpenter's training schedule does not suggest added emphasis on window or periscope navigation training in comparison to Glenn.³⁴ The post-flight training analysis for MA-7 notes: "[the astronaut] was less well prepared for some activities which could not be properly simulated and practiced before flight, such as the gyro-re-alinement and the more extensive attitude-change maneuvers."³⁵ The analysis goes on: "In addition, it should be noted that the horizon-scanner malfunction encountered during the flight (a part of the ASCS) could not be simulated on the procedures trainer, nor could practice be given in the analysis of instrument reference problems because of the lack of a system for simulating the view through the spacecraft window."³⁶

It appears the experience gained during Glenn's MA-6 Flight that yaw determination was not so good came too late for effective action by the MA-7 Flight of Carpenter only three months later. Although, Glenn's flight plan contained a controlled precise yaw maneuver early in the orbital phase, Carpenter's two yaw maneuvers were for photography only not evaluation or learning purposes. Glenn's experience that there was a "learning period" involved in determining yaw suggests the introduction of a precise yaw familiarization maneuver early in Carpenter's flight was advisable, but not included in the flight plan.³⁷ The lack of significant attitude control issues on Glenn's flight with a functioning ASCS looks to have contributed to the shift to more experiments on MA-7, instead of treating it as another test flight that needed to explore the navigation issues related to yaw discovered on MA-6.

The possibility of an alignment problem that required precise yaw determination, in a hurry, with reference to the periscope or window, rather than instrumentation displays, was not adequately anticipated, acted upon, or available for simulation and a likely contributing factor in the yaw error experienced by MA-7 on reentry. The further exploration of yaw determination waited until the MA- 8 flight at which time additional simulators were introduced, the issues related to yaw better understood, and improved navigation aids in the form of the reticle available. In the matter of yaw Carpenter's response and performance was a foreseeable consequence of the training and simulation methods available prior to his flight, and a part of the trial and error process in building the learning curve of manned spaceflight.

¹ Responses to the overshoot vary: "The Man Malfunctioned", Kraft, Chris, Flight, My Life in Mission Control, Dutton 2001, p.162; distracted crewman behind in flight plan without intervention from ground control, too many experiments. Kranz, Gene, Failure is Not an Option, Berkley Books NY 2001, p.91; "It was kinda sloppy." Slayton, Donald K, "Deke" and Cassutt, Michael, Deke!, A Forge Book NY 1994, p.114; "Scott got caught up in some scientific experiments and wasn't ready for his retro-fire." Cooper, Gordon, Mercury 7 Astronaut, Leap of Faith, Harper Collins NY 2000, p.35; ". . . the objective data would be ignored. Once it had begun, the denigration of Carpenter had to proceed at any cost." Wolfe, Tom, The Right Stuff, Farrar, Straus and Giroux 1979, P.302; "everyone failed to perceive that Aurora 7's onboard navigation tools were malfunctioning. . . with its malfunctioning gyros, the spacecraft could not have maintained adequate control during retrofire. Mercury Control may have viewed the manually controlled reentry as sloppy, but the spacecraft came back in one piece. . . Aurora 7 provided proof of why it was important for man to fly in space." Carpenter, Scott and Stoever, Kris, For Spacious Skies, Harcort, Inc. 2000, pp.302-0303 (hereinafter Skies)

² Grimwood, James M., Project Mercury A Chronology, NASA SP-4001, 1963, p.56 (hereinafter Chronology)

³ NASA, Mercury Project Summary Including Results of the Fourth Manned Orbital Flight May 15 and 16, 1963, NASA SP-45, Oct 1963, pp.171-172 (hereinafter Summary)

⁴ Chronology, p.56

⁵ Chronology, p.94

⁶ Chronology, p.102

⁷ Chronology, p.117

⁸ Summary, p. 174-175

⁹ Summary, p.184

¹⁰ Summary, p. 173

¹¹ Summary, p.185; Both Shepard and Grissom found time in the ALFA trainer was necessary but Grissom felt the horizon simulation which was only an illuminated band should be improved. NASA Manned Spaceflight Center, Results of the Second US Manned Suborbital Space Flight, July 21, 1961, p.40 (hereinafter Second); Grissom also acknowledged the ALFA trainer was the only training in visual control of the spacecraft. Id, p.48 Grissom felt the rush of events leading up to his mission prevented more time on the ALFA which he used to practice using the earth-sky horizon as his primary means of attitude control. Id. Deke commented on the ALFA and another device referred to as a Link trainer that was setup in the Moorehead Planetarium to simulate navigation by stars through the spacecraft window Deke found very valuable experience. NASA, NIH & NAS, Proceedings of a Conference on Results of the First US Manned Suborbital Space Flight, June 6, 1961, p.55

¹² Summary, p.185

¹³ Summary, p.172

¹⁴ NASA Manned Spacecraft Center, Results of the First United States Manned Orbital Space Flight, February 20, 1962, p.122 (hereinafter First)

¹⁵ Id.

¹⁶ Id.

¹⁷ Boynton, John H., Manned Spacecarft Center Houston Texas, NASA First United States Manned Three-Pass Orbital Mission (Mercury-Atlas 6, Spacecraft 13), Part I - Description and Performance Analysis, March 1964, p.134 (hereinafter Analysis First)

¹⁸ Id.

¹⁹ NASA, Manned Spacecraft Center, Postlaunch Memorandum Report for Mercury-Atlas No. 6 (MA-6), Part I - Mission Analysis, March 5, 1962, p.11-4 (hereinafter Memorandum No. 6)

²⁰ NASA Manned Spacecraft Center Project Mercury, Results of the Second US Manned Orbital Space Flight, May 24, 1962, p.70 (hereinafter Second)

²¹ Id.

²² NASA Manned Spaceflight Center, Cape Canaveral, Florida, Postlaunch Memorandum Report for Mercury- Atlas No. 7 (MA-7), Part I - Mission Analysis, June 15, 1962, p.7-13 (hereinafter Memorandum No. 7)

²³ Summary, p.284

²⁴ NASA Manned Spacecraft Center Project Mercury, Results of the Third United States Manned Orbital Space Flight, October 3, 1962, p. 42 (hereinafter Third)

²⁵ Third, p.44

²⁶ Id., and Summary, p.285

²⁷ Third, p.44

²⁸ Id.

²⁹ Summary, p.285

³⁰ McDonnell Aircraft Corporation, Project Mercury Familiarization Manual, NASA Manned Satellite Spacecraft One Day Mission, 1 December 1962, p.13-6

³¹ Summary, p.285

³² Memorandum No. 7, p.7-19

³³ Analysis First, p.142

³⁴ Id., at p.133; Grissom stated after his MR-4 suborbital flight in Liberty Bell 7 that he felt additional practice in the ALFA trainer using the window reference would have been desirable and more time at the planetarium and for map study. Second, p.40 and P.48; Glenn's postflight debriefing comments on the other hand included the observation that the ALFA trainer was less valuable than the procedures trainer for attitude control familiarization.

³⁵ Boynton, John H., NASA Manned Spacecraft Center Houston, Texas, Second United States Manned Three- Pass Orbital Mission (Mercury-Atlas 7, Spacecraft 18), Description and Performance Analysis, p.14 (hereinafter Analysis Second)

³⁶ Id.

³⁷ "A thoroughgoing attitude check, during the first orbit, would probably helped to diagnose the persistent, intermittent, and constantly varying malfunction of the pitch horizon scanner." Skies, p. 283

Click image for PDF Version

Click image for PDF Version of Alan Shepard's "Training by Simulation"