Matrix Management Creates Columbia Demise And
So Much More
by Andy Turudic
Many of us have worked in matrix-managed organizations in the past or are currently functioning within one. Unlike the ancient, tried-and-true master/apprentice approach, where experience is slowly gained and broadened under a mentor with ultimate asset and task responsibility, the matrix approach assumes that those most suited to an engineering task should remain where they are, as specialists. Under matrix management, Program Management and Systems Engineering is better suited to the seasoned and extroverted pilots of Powerpoint, Excel, and Project than it is for the humble person who successfully designed 30% of the circuit packs in an advanced packet switching system. The critical decisions affecting costs, reliability, and schedule are traded off in the clouds, with the agenda of clearing the slate for the next program that's coming along, using, of course, "the designer that did our previous program's cards successfully." Nothing exemplifies this system's pitfalls more than the recent report on the Columbia Orbiter disaster. If you think the report and recommendations are exclusive to NASA's problems, blame yourself for not being enlightened by others' errors the next time your program goes down in flames for not listening to humble engineers.
NASA has a classically-matrixed organization that has Program Management with matrixed access to specialists in numerous areas of engineering. The program managers are primarily concerned with the classical aspects of schedule, cost-containment, and of ensuring that the next program was not affected by any schedule slip of the previous one. The engineering organizations were concerned with the purity of engineering -- doing what's right, no matter the cost in pecuniary, temporal, or consequential terms.
In Columbia's case the engineers were concerned about a two by three foot tumbling wedge of foam striking the shuttle at a relative speed of over 500 mph and they immediately began performing simulations and photographic analysis. Despite having had previous non-disabling strikes on shuttles by debris at least half the volume/mass, the only simulation tool available to the engineers was micro-meteoroid simulation software that typically fired pea-sized virtual pellets at the leading-edge carbon-carbon material to gauge the effects on the material and its ability to protect the thin underlying aluminum structure. While preliminary simulations indicated that there would likely not be a problem, the engineers, as engineers do, placed caveats that the tool was being used outside the boundaries for which it was designed when they gave their first back-of-the-envelope assessment to Program Management.
Program Management heard what they wanted to hear -- the simulation said "no problem" and the engineers' offers to perform a more rigorous set of analyses was declined, particularly since the PMs appeared to not want people coming in during the Martin Luther King holiday. The Inquiry's report is silent regarding whether overtime was to be paid if people worked the long weekend. Program Management was also well aware of a foam strike on a previous shuttle mission, stating that the foam was of the same density, and that the prior foam strike had posed no threat to the mission and had only affected the turnaround time of the ship for its next mission. To achieve a closure objective in their minds, this was a familiar foam strike, simulations said there wasn't a problem, and Program Management had previous anecdotal evidence that a ship came back successfully, ignoring the fact that this piece was twice as large and that any high-schooler can predict contained at least twice the energy due to its increased mass, not to mention the plate area combined with extremely large dynamic pressures of flight at Mach 2.49 at the instant of impact.
Program Management also, quite glibly and explicitly, stated to engineering that if there was a problem, there was nothing they could do about it anyway. No brainstorming was ever invited, though engineers did exchange e-mails on what-if scenaria, including landing with blown out tires and estimating that the crew would survive the resulting slap-down of the cabin onto the tarmac. PM's death sentence is contradicted by the Board of Inquiry which determined it was possible to commission a rescue/repair effort within two weeks using the shuttle Atlantis. Again, a rescue would have meant a severe delay in delivering a large and significant component to the International Space Station which appears to be destined to be NASA's cash cow for commercial experimentation. Program Management appears to have understood that heads were going to roll if Atlantis didn't deliver on time, whether rescue-distracted or grounded by shedding foam. Ironically, the last shuttle-based orbital science work before the International Space Station would take over completely was the last flight of Columbia -- the only shuttle with the ability to house a science module in its payload bay.
A motivated Program Management, instead of neutral polling for remaining concerns from the engineers in meetings, used classical bullying tactics that work on engineers by stating the conclusion and a superficial rationale that justified the conclusion through a paraphrase of engineering's words, without caveat, and then bullying consensus by peer exposure: A "team" decision that foam was not and is not a concern. This kept Atlantis on schedule and wrote Columbia's epitaph. It was "just a flesh wound" incurred in the search of the Holy Grail of knowledge and commercial gain. Later in testimony to the Board of Inquiry, Program Management stated that no one person should be blamed for the accident; it was a team decision.
Some engineers appear to have sensed that they were not going to be able to assess the actual shuttle situation through the authorization of the program managers since PM had achieved the desired closure. Program Management openly stated that there were onboard experiments to complete and customers that were intolerant of any slip in the schedule for mission return. One engineer took it upon himself to contact the Air Force's Cheyenne Mountain facility in Colorado to get the ball rolling on having high-resolution ground-, or space-, based photos taken of the shuttle in orbit to view any damage. Though the report doesn't spell it out, this engineer appears to have understood that the proper assessment of the damage on the Shuttle, if any, would not be authorized by the bureaucrats, and that due diligence needed to be done. It was easier in his mind to beg forgiveness than get permission.
The Air Force appeared to comply with the request and began its investigation as to the appropriate asset for imaging Columbia. Likely if from space, this would mean a fuel burn to get the satellite into a parallel orbit with the shuttle. In Program Management's view, a satellite or ground-based telescope photo may have exposed them to aligning the shuttle to the satellite, cancelling the mission since fuel, normally used to run fuel cells, would be burned off manouevering for a view and then back for a landing. This aspect again is not mentioned in the report and would likely have resulted in NASA's inability to collect on the commercially-contracted experiments that were incomplete or uninitiated.
In parallel to the renegade request, the imaging group at NASA, a group that was chartered to obtain and interpret high resolution images of the shuttle also initiated a request with the Air Force. The renegade engineer, as most all engineers would do, told top Program Management about what he had initiated, after he had done so. Such behavior, of course, is organizationally disruptive in a matrix, prompting immediate phone calls from the top to ascertain that no-one had required such photos, at which point the photos were cancelled, likely as a demonstration of the power of office. At NASA, as with most matrixed organizations, the master/apprentice approach of begging forgiveness doesn't work -- forgiveness must come from the matrix, which is next to impossible. To further ensure the powers of office were respected a phone call was made by the NASA liaison, another matrix management specialist function, to the Air Force that they were not to initiate any requests from NASA without his authorization.
Sadly, we have spent over $300 M in the debris recovery effort, the equivalent
of two shuttle launches. Were we not preparing to attack Iraq I can't help
but wonder if the Air Force would have imaged Columbia out of curiosity
or national service to allow Atlantis to rescue a national asset and to
save lives. Instead we find ourselves with nine dead patriots as a direct
result of the accident and with reassigned bureacrats that still hold their
jobs permitting them to bully engineers in yet another program. Now I only
wish there was an e-mail address in the report so I could suggest what may
be the problem with the foam before political momentum builds to spend more
money on a new generation of shuttles with new unknowns. It seems that matrix
management favors the upper management that chooses its use, and only confuses
the creative and curious masters and apprentices of a craft.
Contact the author