* Ignition switch flaw linked to at least 13 deaths
* Congress probes automaker's failure to fix until 2014
* NHTSA says GM "slow to act" on defects and recalls (Adds document detail, recasts top)
By Eric Beech, Paul Lienert and Richard Cowan
WASHINGTON/DETROIT, April 11 (Reuters) - General Motors engineers were well aware of serious problems with ignition switches in GM small cars, but rejected several opportunities to make fixes, according to dozens of confidential documents released on Friday by a Congressional committee investigating the deadly defect.
Parts supplier Delphi Automotive also repeatedly tested switches and found they did not meet GM specifications, according to emails and other memos.
The internal documents from GM, Delphi and a U.S. safety agency chart numerous examples of switch failure, of the sort that led GM earlier this year to recall 2.6 million cars to replace defective switches now linked to at least 13 deaths.
The documents, the first tranche of some 250,000 pages, were released by the House Committee on Energy and Commerce, which last week grilled GM Chief Executive Mary Barra on the automaker's slow response to problems that GM first documented in 2001.
Committee Chairman Fred Upton, a Michigan Republican, said the documents illustrate "failures within the system." Other lawmakers have questioned whether GM's action are criminal.
Meanwhile, a top official with the National Highway Traffic Safety Administration told General Motors in a July 2013 email that the automaker was "slow to communicate, slow to act" on defects and recalls.
Still to be answered is whether top GM executives were aware of the issues early on, as engineers struggled to pinpoint causes and solutions for ignition switches that could be turned off inadvertently with the vehicle in motion, causing the engine to stall and cutting power to steering, brakes and airbags.
GM says it is cooperating with Congress and conducting its own "unsparing" investigation of the circumstances that led to the recall.
The documents show the automaker repeatedly elected not to fix or replace the faulty switches, because there was no acceptable "business case", an indication the solution was deemed too expensive.
Federal regulators as early as 2007 were concerned that GM was dragging its heels on safety measures as consumer complaints mounted, but top officials at NHTSA never followed through on staffers' recommendations to open a broad investigation, according to the documents.
EARLY EVIDENCE OF PROBLEM
It was determined eventually by GM that the switches didn't have enough torque, the rotational force required to keep them from moving from the "run" position to "accessory" which shut down the engine.
A root cause of the problem was a tiny set of parts, called a detent plunger and spring, that helped keep the ignition key in position while the car was running.
A GM engineer at the automaker's mid-Michigan test track encountered an early problem with the switch while driving a prototype of the 2003 Saturn Ion in July 2001.
An internal GM memo on the incident noted that a "tear down evaluation on the switch revealed two cause of failure. Low contact force and low detent plunger force." The memo said both issues were resolved with newer parts, and the case was closed that November, less than a year before the all-new Ion went into production. One of the engineers who signed off on the fix was Ray DeGiorgio, the designer of the switch.
DeGiorgio was one of two GM engineers placed on paid leave earlier this week as GM continues an internal investigation of the recall. GM did not explain the move and DeGiorgio could not be reached for comment.
Supplier Delphi submitted a batch of Ion switches in December 2001, but informed GM that its tests showed many of the switches did not meet GM's torque specifications, according to GM validation documents. Nevertheless, GM approved the parts for production in May 2002, another document shows. The first Ion rolled off the line in August that year.
GM was already beginning to monitor customer complaints about engines stalling in the Ion in 2003, but noted "technicians are rarely able to duplicate the concern."
FAULTY COBALT SWITCHES
In the meantime, the automaker had begun to develop a sibling to the Ion, the 2005 Chevrolet Cobalt. While testing some of the first cars off the assembly line in October 2004, engineer Gary Altman noted that "the driver's knee bumped the key in such a manner as to turn off the ignition."
Altman, the program engineering manager for the Cobalt and Ion, was the second GM engineer put on paid leave this week for undisclosed reasons. He could not be reached for comment.
Engineers considered possible remedies, but it was decided that "the tooling cost and piece price are too high" and the lead time required to make the change too long. The case was closed in March 2005, with engineer Blendi Sullaj noting, "None of the solutions represents an acceptable business case."
Problems persisted with engines stalling in the Cobalt, which GM engineers by then had traced to the faulty switches.
In another document, Delphi engineers on June 14, 2005, discussed a request from GM that they perform an analysis on the switch.
"Cobalt is blowing up in their face in regards to turning off with the drivers knee," wrote Delphi engineer John Coniff.
A second GM investigation in June 2005 for stalling engines in the Cobalt resulted in a recommendation to provide an ignition key insert to customers, but no change in the faulty switch because the "business case (was) not supported" to redesign the part.
Another group of GM engineers in September 2005, in an email chain, discussed postponement until fall 2008 of a proposal to implement a new switch on the Cobalt, Ion and companion vehicles, because the change would add $400,000 in retooling cost, plus an additional 90 cents per vehicle.
Lori Queen, the top executive overseeing GM's small car team, challenged the delay, saying, "I'm not sure it's ok to wait." She could not be reached for comment.
DeGiorgio began working with Delphi engineers to implement a more modest change to the existing switch, according to several GM and Delphi documents.
DeGiorgio signed off on a redesigned switch without a new part number, according to a Delphi internal memo dated May 27, 2006. Since the redesigned switch, which was installed that fall in 2007 Cobalts and Ions, carried over the old part number, it was harder for GM investigators in later years to trace and pinpoint problems in the cars.
DeGiorgio, in a 2013 deposition in a Georgia lawsuit against GM, denied any knowledge of or involvement in the 2006 design change on the switch.
With "much left to examine," according to Upton, the House panel, as well as a Senate panel conducting its own investigation, is expected to take weeks examining the load of materials collected from GM, Delphi and NHTSA.
Both panels held hearings last week. Additional hearings are expected later this spring or into summer, when Barra and other GM executives are expected to testify.
According to one document obtained by the committee, Barra, who assumed the top position in January, received an email in 2011 pointing to steering problems in GM models that later were recalled.
That email cited a New York Times story dated Oct. 3, 2011, which reported on NHTSA deliberations concerning Saturn Ions and Chevrolet Cobalts that were experiencing steering problems related to a loss of power.
The email to Barra, however, does not mention ignition switch problems, something the new CEO said she became aware of just last December. GM said the two issues were "completely separate".
Congress is trying to determine whether GM officials failed to react in a timely way to the critical safety defect and whether NHTSA regulators also may have failed to carry out their duties.
Frank Borris, head of NHTSA's Office of Defects Investigation, said in a July 2013 email to GM executive Carmen Benavides that the company was more difficult to work with than other automakers and he cited six instances in which the agency disagreed with GM on safety issues. This was the same email that accused GM of being "slow to communicate" and "slow to act" on details and recalls.
Some members of Congress already are discussing the possibility of passing legislation to increase civil and criminal penalties for automakers' failure to react quickly to safety concerns and to tighten reporting requirements in crashes involving fatalities.
(Additional reporting by Julia Edwards and Marilyn Thompson in Washington, Bernie Woodall in Detroit, Jessica Dye in New York, and Peter Henderson in San Francisco; editing by Karey Van Hall and Matthew Lewis)