By Marilyn Thompson and Paul Lienert
WASHINGTON/DETROIT, March 30 (Reuters) - Among the most critical unanswered questions in General Motors' mounting crisis over a defective ignition switch that led to crashes and at least 13 deaths is who, eight years ago, approved a change in the part for new vehicles and why that didn't lead to a recall of older cars to fix the problem.
Among the evidence that investigators inside and outside of the company will need to reconcile is the company's timeline of the switch's engineering, which describes an unnamed GM engineer approving the design change, and the deposition of a senior switch engineer, Ray DeGiorgio, who said he was not aware of it.
GM declined to discuss the matter beyond its timeline or to make DeGiorgio available for an interview. DeGiorgio referred Reuters to GM, which has presented no evidence to suggest that he was the executive approving the design change. Delphi Automotive, which made the part for GM, also declined to comment.
However, Delphi told U.S. congressional investigators last week that GM approved the original part in 2002, despite the fact it did not meet GM specifications, according to congressional aides on Sunday.
GM CEO Mary Barra is due to testify to Congress this week on the recall. Lawmakers have requested documents from GM and could call other executives for further hearings.
The availability of the fix in 2006 raises the question of whether the no 1 U.S. automaker could have recalled older models sooner and saved lives.
GM on Friday, almost eight years later, increased the recall by nearly one million to 2.6 million vehicles, spanning models years between 2003 and 2011. It has also increased the confirmed fatalities to 13 from 12.
GM's timeline, as prepared for federal regulators, says that "the GM design engineer responsible for the ignition switch" in six recalled vehicles approved an ignition switch design change on April 26, 2006.
The change was made to the portion of the switch that holds the ignition key in place as it clicks between off, accessory and on positions, called the detent plunger and spring.
GM switches in Chevrolet Cobalts and other models were prone to being jostled into accessory mode while cars were moving, often by something as small as additional keys hanging off a key ring. That would shut off engines and disable power steering, power brakes and airbags, which led to many crashes. Pressure from the new, tighter spring on a slightly longer plunger resisted bumps and keeps the engine running.
"We certainly did not approve a detent plunger design change," DeGiorgio said last April in a deposition taken by plaintiff's attorney Lance Cooper. DeGiorgio identified himself as the "main responsible engineer" for the ignition switch in the Cobalt and the Saturn Ion, which is also part of the recall.
Cooper's clients, the family of Brooke Melton, sued GM after the young woman died in the 2010 crash of a 2005 Chevy Cobalt. They settled with GM.
DeGiorgio recalled GM bringing ignition switch design in-house around the time the 2005 model Cobalt was being developed, but it still worked closely with its supplier, Delphi Automotive, which made the switch. Suppliers alert GM to any changes and then must test and validate the redesigned part.
"Revalidation would mean that you have to go through the same procedures that we had done previously," he said. "My point here is changing the spring would have required a revalidation by the supplier for the new spring," he said. "That was never brought to my attention."
When asked if anyone else at GM was aware of the change GM adopted in 2007 models, he said, "I am not aware about this change."
In the deposition, DeGiorgio questioned whether the old switch was dangerous. The weekend before he testified he drove his son's 2007 Chevrolet Cobalt around their neighborhood to try to replicate conditions that contributed to the accidents.
In the moving car, he coaxed the ignition switch into the "accessory" position, disabling power assist systems. He had no trouble safely maneuvering his son's car to the side of the road, he said.
"You definitely can control the car ... That's not an issue," he testified, reflecting GM's official position until the recall.
Attorney Cooper only became aware of the ignition change after hiring an engineering consultant, Mark Hood, who found the change in the spring.
Neither GM nor Delphi changed the part number when they redesigned the switch, GM has said. Experts say that is contrary to industry practice and made it all the more difficult for investigators including Hood to understand what originally went wrong.
Retired GM executives familiar with the automaker's longstanding engineering, manufacturing and purchasing processes say that even minor changes in a car's basic components must be thoroughly documented.
Internal records should show if the changed ignition switch was tested and validated by GM engineers, they say.
"In the auto world, with safety and health issues involved, you can't willy-nilly just be making changes without a paper trail," said R. Bick Lesser, who previously worked as an engineer at Packard Electric, which became Delphi-Packard.
Packard was a GM supplier for wire harness electrical systems.
"If it is not there, this would be a total breakdown of the system," said Lesser, who wrote a book on the division's relationship with GM.
Congressional investigators are still gathering records from the companies and say they are prepared to probe the company's manufacturing controls.
"It is critical that all potential problems are identified and fixed as soon as possible so we can keep Americans safe," said Representative Fred Upton, a member of the House energy and commerce subcommittee that will hear testimony from Barra on Tuesday.
(Additional reporting by Eric Beech, Richard Cowan and Julia Edwards in Washington, Ben Klayman in Detroit and Jessica Dye in New York; editing by Peter Henderson and Frances Kerry)