On March 20, 2000, approximately 0731 mountain standard time, a Dornier Luftfahrt GmbH 328-100, N329MX, operated by Air Wisconsin, Inc., and doing business as United Express flight 7128, was substantially damaged when the right main and nose landing gears retracted during landing at Denver International Airport, Denver, Colorado. There were no injuries to the airline transport certificated captain and first officer, one flight attendant, and 21 passengers. Visual meteorological conditions prevailed, and an IFR flight plan had been filed for the scheduled domestic passenger flight being operated under Title 14 CFR Part 121. The flight originated at Bismarck, North Dakota, approximately 0534.

According to the flight crew, during the approach they confirmed the landing gear was down and locked. They broke out of the overcast some distance from the runway and had good visibility, so both pilots were looking outside. The first officer landed the airplane on runway 35L in a slight left crosswind. When the airplane touched down, the landing gear warning horn sounded. The right main and nose landing gears retracted. The airplane slid to a halt on the right side of the runway.


The captain, age 37, held an airline transport pilot certificate, dated December 16, 1998, with an airplane multiengine land rating; commercial privileges, airplane single engine land, and a type rating in the Dornier DO-328. He also held a flight instructor certificate, dated July 22, 1999, with airplane single, multiengine, and instrument ratings, and a ground instructor certificate, dated November 7, 1994, with an advanced rating. His first class airman medical certificate, dated October 5, 1999, contained the restriction, "Holder shall wear corrective lenses while exercising the privileges of his airman certificate." His last biennial flight review or equivalent (proficiency check) was dated January 13, 2000, and was accomplished in the Dornier 328-100.

The first officer, age 34, held an airline transport pilot certificate, dated December 4, 2000, with an airplane multiengine land rating; commercial privileges, airplane single engine land, and type ratings in the Dornier DO-328 and Shorts SD3. He also held an expired flight instructor certificate, dated June 12, 1996, with airplane single engine land and instrument ratings, and a ground instructor certificate, dated February 2, 1996, with an advanced rating. His first class airman medical certificate, dated September 1, 1999, contained no restrictions or limitations.

According to Air Wisconsin records, the flight attendant was certified to serve on both the Dornier 328-100 and the British Aerospace BAe 146.


N329MX, a Dornier Luftfahrt GmbH 328-100 (s/n 3049), was equipped with two Pratt & Whitney PW119C turboprop engines (s/n PCEAB0010, PCE116051), each rated at 2,180 horsepower. The engines had accrued 7,265:14 and 9,070:02 hours of operation, respectively, at the time of the accident. The airplane was being maintained under a continuous airworthiness inspection program, and had accrued 7,596:18 hours at the time of the accident.


The following weather observation was recorded at Denver International Airport 1 minute after the accident:

Wind, 350 degrees at 19 knots, gusts to 28 knots; visibility, 9 s.m.; ceiling, 1400 feet overcast; temperature, 28 degrees F.; dew point, 21 degrees F.; altimeter setting, 29.74 inches of mercury.


Denver International Airport is at an elevation of 5,431 feet msl. Runway 35L is 12,000 feet long and 150 feet wide. It is concrete and grooved. It is served by an ALSF-2 approach light system, precision approach path indicator, high intensity runway lights, and runway centerline lights.


The airplane was equipped with a Fairchild A-100A cockpit voice recorder (CVR) and a digital flight data recorder (DFDR). Data from both recorders were downloaded at NTSB headquarters on March 23, 2000.

According to the CVR transcript, the first officer requested that the landing gear be lowered at 0729:34. Two seconds later, a sound similar to the landing gear being operated was recorded. At 0730:28, the captain reported "gear down, three green," to which the first officer replied, "down, three green, I see it." At 0730:43, the captain reported, "gear down, three green...before landing checklist is complete." At 0731:53, a sound similar to a decrease in engine rpm was recorded and the captain remarked, "Nice." At the same time, a sound similar to the landing gear warning horn started. The captain said, "Oh, *," and the first officer asked, "What the #'s that?" The captain replied, "#, I don't know. At 0731:59, the sound of a thump was recorded and the captain said, "Hang on." The captain reported to the control tower that the gear had collapsed at 0732:03. At 0732:08, the sounds of impact and grinding noises started and continued to 0732:29. At 0733:00, after the passengers had been evacuated, the captain asked, "We had three green, didn't we?" The first officer replied, "We did. We called for the..." The captain then said, "Don't ask me, dude. I have no idea. I should have had you go around." The first officer said, "We touched down but..." This was the end of pertinent conversation.

Some of the findings of the DFDR group chairman's factual report were:

"(1) At 0729:44, while descending through 7,000 feet pressure altitude, the airplane's landing gear discrete changed state from 'up' to 'down and locked' (the landing gear was extended and indicated down and locked, the source data being the position of the nose landing gear);

"(2) At 0731:17, the autopilot engage discrete changed from 'engaged' to 'not engaged' (the autopilot was disengaged);

"(3) At 0731:50, vertical acceleration was recorded at a local maximum of 1.080 g's. At 0731:51, the landing gear discrete changed state from 'down and locked' to 'up' while at an airspeed of 120.0 knots (the nose landing gear became "unlocked," or unsafe, about 2 seconds after the radio altitude reached 0 feet);

"(4) During the approach, hydraulic pressure was recorded at 3,024-3,029 pounds per square inch (psi). At 0731:51, hydraulic pressure was 3,009 psi, dropped to a local minimum of 2,683 psi at 0731:53, then 3,024 psi at 0731:55.

"(5) At 0731:51, roll attitude was recorded at a local minimum of -2.37 degrees (left wing down). Roll values increased to about zero at 0731:55 and then increased to a local maximum of 4.83 degrees (right wing down) at 0731:57, decreased to another local minimum of -0.26 degrees at 0731:59, then increased and remained at or about a maximum of 18.02 degrees at 0732:27 through the end of the recorded data (uncommanded right main landing gear retraction).

"(6) The left and right engine torque and propeller speed values approximated each other during the descent and landing. At 0732:07, however, the left and right engine torque values diverged, 7.53% and 11.59% left and right engine respectively. [According to Fairchild Dornier, flight idle is 12% which "is enough to produce considerable lift, thus reducing the load on the landing gear and WOW (weight on wheels) onset."] Over the next several seconds the left engine values continued a gradual increase whereas the right engine values increased to a local maximum of 33.12% at 0732:11 and then decreased to zero at 0732:16. The left and right engine propeller speeds were approximately equivalent until 0732:10 when the values diverged. The right propeller speed values dropped to zero at 0732:16.


The airplane was examined in a hangar after it had been towed off the runway. Damage was confined to the right side skin and bulkheads, aft of the copilot's station and forward of the wing.

Examination of the cockpit shortly after the incident disclosed the landing gear control lever was in the DOWN position, and the landing gear status lights indicated the left main landing gear to be DOWN AND LOCKED (green) and the right main and nose landing gears to be UNSAFE (red). After the airplane was jacked up and the right main and nose landing gears were lowered and pinned in place, it was towed back to a hangar where initial tests were conducted. The airplane was placed on jacks and the landing gear was cycled 30 times without difficulty. The up and down locks were undamaged and moved freely. The proximity switches tested satisfactorily. The landing gear stopped once a WOW (weight on wheels) condition was simulated.

The landing gear control lever and lever lock solenoid were tested and found to operate normally. Attempts were made to get the handle stuck in the middle position, but the over-center lever prevented this. The handle could not be forced past the lever without going into manual override.

The landing gear selector valve was undamaged and operated normally.

The hydraulic pressure-sensing valve was unremarkable. The hydraulic system was full of fluid and none of the filters had been bypassed.

Several gear extensions were made utilizing the emergency gear extension system. No anomalies were noted.

The proximity switch electronics unit (PSEU) tested normally. A complete PSEU IBIT was performed. All systems tests were satisfactory. The PSEU nonvolatile memory (NVM) was downloaded and revealed numerous faults, none of which were relevant to the accident.


On April 12, 2000, the PSEU (p/n 8-700-04, s/n 77) and the right main landing gear harness and proximity switches were tested at the facilities of Eldec Corporation in Lynnwood, Washington. The PSEU tested satisfactorily. It was then subjected to temperature extremes (cooled to -40 degrees C. for 2 hours, then heated to 65 degrees C. for 1.5 hours, then returned to ambient temperature). Again, it tested satisfactorily. The harness and sensors tested satisfactorily.

On May 10, 2000, the landing gear selector valve (p/n HP848800-4, s/n 1487), hydraulic pressure sensing valve (p/n QA07376-01, s/n RC22515), and the landing gear control assembly (p/n 717801-1, s/n 118) were tested at the facilities of Fairchild-Dornier in San Antonio, Texas. The landing gear selector valve functioned normally. It was then partially disassembled and examined. No anomalies were noted. The pressure-sensing valve functioned normally but was slightly below maximum flow. Continuity checks through all switch positions were then performed on the landing gear control lever. No anomalies were noted.

On August 10, 2000, the landing gear control assembly was subjected to more elaborate tests at the facilities of the Eaton Corporation in Costa Mesa, California. No pertinent faults were noted.

Hydraulic fluid samples drawn from the filter pack, right main landing gear down lock assist actuator, and the return line for the landing gear selector valve were analyzed at Aviation Laboratories, Kenner, Louisiana. According to their report, the samples were normal for wear metal, particle size distribution, viscosity, total acid number, Karl Fisher water, and total chlorinated solvents by volume. An insufficient fluid quantity precluded a particle count and viscosity analysis of the sample taken from the filter pack. A German military laboratory, WEWIB, made an infrared spectrometer analysis of the sample and found it contained an insignificant amount of MIL-PRF-83282D.

Partial systemic wiring tests were made while the airplane was undergoing repairs. No anomalies were noted.

The integrated avionic computer system (IAC), which provides a readout capability of systems status, was checked and no relevant faults were noted.


Fairchild Dornier provided the following systems information: The gear safe/green light indication is dependent on each individual downlock sensor, and is not influenced by WOW signals. The PSEU logic for illuminating the gear unsafe/red light would require a retract command and each uplock would have to be unlocked. Once a WOW signal is sensed, the ground dislock in the gear lever is activated, preventing landing gear retraction by inadvertently raising the landing gear handle. It can only be moved by manually overriding the dislock solenoid by pushing a button on the landing gear selector panel.

There are three microswitches in the landing gear control unit that provide gear position input: one to the extend side of the landing gear selector valve solenoid (28 VDC), one to the retract side (28 VDC), and one providing input to the PSEU. In order for the landing gear to retract, there must be an input from the PSEU and an input from the landing gear selector panel, and both inputs must be in agreement in order for the circuit to close. Output from the PSEU enables or inhibits gear retraction (DO41). DO41 is processed in the PSEU depending on the microswitch that provides gear lever position information to the PSEU (retract command input) in conjunction with WOW sensor inputs. The retract command signal-ground is enabled only by the PSEU if at least one of both WOW switches of all three gears indicates WofW (weight off wheels). DO41 uses its own logic, and will allow gear retraction when at least one sensor of each landing gear indicates WofW.

The WOW signal is derived from proximity switches that are in a NEAR position if there is no load on the landing gear. In order to get a FAR signal, there has to be some movement of the shock absorber caused by a load on the landing gear (or WOW). When the left main landing gear proximity switches senses WOW, the PSEU removes the ground signal from the retract solenoid of the landing gear selector valve, thus stopping gear retraction.

In addition to the Federal Aviation Administration, parties to the investigation included Dornier, Fairchild-Dornier, Eldec Corporation, Air Wisconsin, Inc., Messier Services, and the Air Line Pilots Association.

Use your browsers 'back' function to return to synopsis
Return to Query Page