Hey: That CRM Stuff Really Works!
by Major Bernard Mater, USAF
It seems that when we either receive Crew Resource Management (CRM) training or go through the annual refresher, we hear nothing but war stories about aircrews who died because they did not use CRM. I would like to relate an "almost incident" experienced by my crew and how CRM ended up saving our lives and a USAF C-130H aircraft.
It was a beautiful Alaskan day for flying -- unlimited visibility, ceilings about 8,000 feet, with a little light rime icing and light-to-moderate turbulence forecast in the Anchorage bowl area. However, the weather forecast for the area north of the Alaskan Range (where we were going to fly) was CAVU (ceiling and visibility unlimited). Our mission was to fly from Elmendorf to Eielson, then from Eielson to two long-range radar sites -- Cape Lizburne and Sparrevohn AFSs. If you look at these two sites in the Airfield Suitability and Restrictions Report (ASRR), you will see that they are both special certification airfields. They require an unusually high degree of preflight planning, aircrew coordination, and aircrew experience levels. The pilot in command (PIC) was a fairly new aircraft commander, the copilot was an evaluator pilot (CP/EP), and the flight engineer (FE), navigator (myself), and loadmaster (LM) were all instructors. We had a flight surgeon on board (flying for proficiency) and a second instructor pilot (IP), who was receiving qualification training for Sparrevohn AFS. We joked in base ops that with all this experience, "We were a Class A waiting to happen!"
The preflight was largely uneventful, though there were minor discrepancies that maintenance quickly fixed. During the taxi to the runway, we noticed a slight burning rubber smell shortly after the FE turned on the cockpit air conditioner. The engineer turned it off and the smell went away. After a quick scan of the cockpit and under the flight deck, he turned the cockpit A/C back on, and the smell did not return. We decided it was either a CND ("can not duplicate") or more likely just some dust in the A/C system.
During the takeoff brief, the PIC told the crew that if during the takeoff, before the "GO" call, there was something wrong, the terminology was "REJECT" and state whether it was a "prop, engine, system, or load malfunction." All of that is part of our standard cockpit operating procedure.
During the lineup check, the pilots and engineer noticed the torque on the number two engine was lower than the other three engines, but still within limits. After takeoff, the engineer analyzed the number two engine performance and concluded we had a 95% engine. (This means that it was performing above the low end of the acceptable spectrum.) However, the pilots and engineer decided to perform a "max" power check after landing at Eielson to see how well the engine was working. This would be crucial since we would be using assault landing and takeoff procedures at Cape Lizburne and Sparravohn.
When it was time for takeoff at Eielson, we performed the max power check. With all bleed air off, all engines were performing like 100% engines -- including number two. Since we were going to turn the bleeds off for our landings and takeoffs at the next two sites, we decided we were "good to go!"
During climb out, I commented about how slowly the aircraft was climbing out -- how it was performing like the "E" models the second IP and I had flown together previously at Yokota AB, Japan. The PIC commented that he noticed the number three engine seemed to have low power, and thought that was what the copilot was looking at during the takeoff at Eielson. The CP stated, "No, it was the number two engine -- it seemed to have lower than predicted torque during the takeoff," -- but he wasn't sure.
The CP went on to explain that number two was fairly close to the number three engine and he elected not to call "REJECT, number two engine." The three pilots and engineer discussed the two warnings in the Dash 1 about normal takeoffs. The first warning says: Above crossover, if the engine instruments are not similar in fuel flow, TIT, or torque with throttles aligned, a propulsion system malfunction may exist. The second: When take-off power is applied, ensure all engine instruments are similar in torque, TIT, and fuel flow. Since the fuel flows for each of the engines and engine instruments were about equal, neither pilot felt a rejected takeoff was in order for a TIT or propulsion system malfunction. The Dash 1 goes on to say that for either a fuel flow or an engine instrument discrepancy, a Temperature Controlling Check should be performed prior to flight. Well, we were certainly past that point and the Dash 1 didn't provide any guidance about whether flight should be continued. We decided to analyze our situation and come up with our own options.
The engineer felt confident he had calculated the proper minimum torque setting for 970 TIT, but offered to let any other crewmember chase through the performance charts with him to reverify the predicted torque. The CP/EP called back to Eielson tower to request runway temperature to compare with ATIS data received, and to use for the calculation. The second IP and engineer then recomputed the data using this temperature. We were using all available resources -- an extra pilot to verify the numbers and Eielson tower versus ATIS to get the best temperature information -- we thought we were a model CRM aircrew. Using this new temperature from Eielson tower to recalculate the takeoff torque setting, it turned out the torque the CP saw was about 800 inch pounds below the predicted torque, instead of the 200 or 300 pounds we had previously thought.
We decided to level off at 20,000 feet (vice our planned 22,000 feet) in order to help the FE do another power check and a temperature controlling check on the number two engine. He now calculated the number two engine was operating at about 92% efficiency. He also believed this was a problem with the mixture control to temperature datum control system. We, as a crew, discussed what other malfunctions it could be, such as a problem with either the fuel control, mechanical governing system, bleed air system, an engine instrument gauge error or failure, or a thermocouple failure. Now that we had leveled off, almost all of the engine instruments (i.e., fuel flow, RPM, oil pressure, oil temperatures, and engine tempratures) were nearly identical and the number two and three torques were within 800 inch pounds of each other, while the outboard torques were a couple of hundred inch pounds higher. Fortunately, the engine power was symmetrical, so we didn't think we would have any aircraft control problems.
But the flight engineer kept coming back to a problem with the number two engine. He had flown this aircraft a couple of days earlier, and decided it was "just low power -- an old engine." The three pilots basically agreed with the FE. I laughed and remarked that this discussion sounded like that old "strengthening of an idea" we were taught about in CRM.
Nevertheless, we felt fairly confident we had a minor problem with the number two engine, maybe with the fuel control. Since we were going to use assault procedures on our next two takeoffs and landings, and the number two engine worked perfectly with the bleed air off during the max power check, we felt confident we could get in and out of the two sites just fine (especially Sparravohn, which was one-way in, one-way out). The PIC asked if anyone had a problem with that, and if they did, he had no qualms about returning to Elmendorf. While he was saying all the right words, the CP/EP commented about the NDB approach he needed and wanted to do at Cape Lizburne, and the second IP said he felt good about taking the aircraft into the one-way site at Sparravohn in order to "get signed off on that site." Were we "pressing," or expressing confidence that we could deliver the food to the remote sites and pick up our return cargo? The "most conservative response" rule would have had us turn around, but no one demanded to go home. Were we a Class A waiting to happen?
At the heart of CRM is the "QPIDR" model. We used the "QPIDR" model to analyze the situation, and take what we thought were the appropriate actions. The first step, "Question," was done through the CP's recognition of low power on number two. The second step, "Promote Ideas," was used to verify the proper predicted torque, perform an engine efficiency check, use the loadmaster to scan the engines, compare the four engines against each other, and brainstorm ideas of what could really be going on. The third step, "Decide," was used in two ways. First, we decided we had a minor problem with the number two engine. Second, we decided this was a problem that wasn't going to prevent us from going to our two sites. However, it was that fourth step, "Review," that probably saved our lives.
As a navigator, I probably know less about aircraft systems than most pilots, and certainly less than the flight engineer -- or at least I am supposed to. Rather than simply "being a passenger," I have often used my lack of expertise to play "Devil's Advocate" during aircraft malfunctions (not meant as a dig at the other crew positions). So I asked, "If at Cape Lizburne we accomplish a normal power check and find number two has less than predicted torque, are we going to keep it on the ground, like the Dash 1 note says, or takeoff using max effort procedures?" The PIC decided on the max effort procedures, especially considering the fact that it seemed we had successfully taken off once with less than predicted torque. We discussed the wisdom of violating "the bible," and reviewed our MCM 55-130, Volume 1, Chapter 10 (local procedures), which allows us to deviate from Dash 1 guidance "in unusual circumstances."
I then commented that there were some things about the analysis that didn't make sense to me, and asked the pilots and engineer to explain them to me. First, would the amount of low power on number two make the aircraft climb so poorly? They discussed this, and concluded that it shouldn't have. I asked if number two was operating so poorly, why was number three now also now operating at less than 100%? They didn't have an answer for this. I had earlier brought out the idea of doing a bleed air check, but we elected not to. Now, I brought it up again. The CP/EP said, "Why not? It is easy to do and costs us nothing." The C-130 bleed down check involves measuring the length of time it takes for pressure to bleed off the manifold as it decreases from 65 psi to 35 psi. This should take a minimum of ten seconds for our version of the C-130H. The FE tried to pressurize the manifold by going to override with all four engine bleed air systems. However, he could only get 62 psi, not the normal 70 or so psi usually attained in the override position. He thought this unusual, but felt it was probably due to number two not putting out enough power. He then turned off the bleed air, even though it was only 62 psi, and the pressure fell to zero in less than a second! We had a problem -- actually several. First, it was now apparent we had a severe bleed air leak. For those not familiar with the C-130, 600C (over 1,100F) bleed air is pulled off the engine to drive three main systems during flight: the air conditioning system, the leading edge anti-icing system, and the aircraft pressurization system. With the bleed air off, the cabin altitude began a fairly rapid increase. This caused our second problem -- we needed to descend or get on oxygen ASAP! Fortunately, we had a flight surgeon on board to monitor everyone for possible physiological symptoms.
We declared an emergency with Anchorage Center and reversed course while asking for a descent to 10,000 feet. Since we were carrying passengers, we turned on the bleed air system long enough to prevent the cabin pressure from exceeding 10,000 feet while we descended. During the descent, we completed the remainder of the bleed air check by checking each of the engines. We started with number two, since we now thought that part of the bleed air ducting had separated or developed a hole. Reinforcing that belief was the return of the "burning rubber" smell we had experienced during the taxi at Elmendorf some three hours ago. Again we couldn't get the manifold pressurized high enough, and the number two engine bleed air couldn't hold the system pressure. This confirmed our bleed air leak. We had 600C air going somewhere it wasn't supposed to. We elected not to test the remaining three engines (we felt that would be pressing our luck).
Needless to say, this was an event-filled five minutes as we descended to 10,000 feet -- the PIC flying the aircraft; the CP talking to Center, getting clearances and declaring the emergency; the FE analyzing the systems with the second IP; me determining terrain clearances and a route of flight through the mountains, and the LM assisting and ensuring the safety of the passengers. Fortunately our training, both individual and crew training, paid off. However, we weren't finished with CRM. We now had to decide whether to fly to Eielson (with no C-130 maintenance facilities and good weather -- but not the "severe clear" that was forecast) or return to our home base, Elmendorf (and potentially fly through icing, which would necessitate turning the bleed air on in order to get rid of ice on the leading edges of the wings and empennage).
Since you didn't read about us in Blue Four News or hear about us on CNN, you can guess the outcome -- we made it home safely. Nevertheless, we continued to use the QPIDR process each step of the way home, and our successful use of CRM contributed to our safe recovery.
On the post-flight inspection, maintenance determined the cockpit air conditioning bleed air valve had failed. This explained both the initial burning rubber smell and where the bleed air was going -- under the flight deck. The maintainers later told us that if we hadn't shut off the bleed air system when we did, we would have soon had a major fire under our seats.
CRM is designed to advert aircraft accidents. Due to successful CRM, I would count our flight that day as "one mishap that didn't happen."
Copyright © 1996-2005 by Neil C. Krey unless otherwise indicated.
Non-commercial reproduction rights granted if the following notice is included:
"Source: Neil Krey's CRM Developers Forum, http://www.crm-devel.org"