Significant changes should be made before the space shuttle's next launch to eliminate sporadic erosion in the heat-resistant material that protects the nozzle of its booster rocket, according to a 240-page internal report being reviewed by NASA.

In describing the erosion, the report said tests show that the material -- a carbon-and-resin liner slightly more than an inch thick inside the nozzle's aft exit cone -- has on several occasions suffered "fractures . . . so numerous, so large, so closely spaced and so extensive that the integrity of the entire {cone} came into question."

The report enumerates steps to enhance the durability of the dense carbon material that shields the cone from the hot propellant exhaust generated at launch. The cone, about 12.5 feet in diameter, directs exhaust and helps the rocket gain enough thrust to ensure a successful liftoff.

Nowhere does the report, obtained in its final form, describe the problem as potentially catastrophic. Top NASA officials say initial concern over the erosion has been tempered by two recent tests, which resulted in the kind of limited erosion considered normal in flight.

"Some of those changes are fairly major changes, and because of {the recent successful tests} . . . we don't feel a need for a radical departure from what we've been doing," said Royce Mitchell, who as NASA's project manager for the solid rocket motor is reviewing the report, sent to the National Aeronautics and Space Administration about two weeks ago.

But four of the six engineers who have spent hundreds of hours since August studying the erosion at the request of the rocket manufacturer, Morton Thiokol Inc., think that the carbon material poses a risk of "borderline performance" and recommend that changes be made before the shuttle's return to space, scheduled for August.

"Few of us can believe a material as bad as this and so little understood has been allowed to remain," said an engineer who attended a meeting in September when the majority presented preliminary recommendations. "Those recommendations were not made lightly."

Two engineers disagreed in written dissents, disputing the seriousness of the problem, questioning the data used to support the recommendations and arguing that additional study would cause unnecessary delays and increase costs. They supported implementing several less radical changes.

At the same time, one of these dissenters, Howard K. Larson of NASA's Ames Research Center, said in his three-page dissent that the nozzle should not be used after the August flight. "The current nozzle represents 1960s technology in its design, materials, fabrication procedures, hardware, location, atmosphere control, etc., etc.," Larson wrote.

This difference of opinion, common within an agency that must deal with evolving technologies, raises the question that has consumed the space agency since the 1986 Challenger disaster and throughout its two-year effort to return to manned flight: How safe is safe?

There are engineers at Morton Thiokol and NASA who argue that all identifiable problems should be corrected to their technological best before the shuttle flies again. But other engineers say that approach sets up an impossible goal: risk-free flight.

"There are probably thousands of issues like this brought up during the course of the discussion leading to flight," Russell Bardos, NASA director of shuttle propulsion, said of the latest study. "You have to decide how and when to address them . . . . All along the way, there is learning done. In this case, what {this means to NASA} is that we don't necessarily need a change in design, but better controls on what we're doing."

The nozzle assembly of the booster rocket has been scrutinized intensely by NASA in preparation for the next launch. The first post-Challenger launch was scheduled for February but has been twice postponed because tests and routine inspections revealed design and hardware problems.

The material in question, carbon cloth phenolic, has been used in every space shuttle flight. NASA's concern about the material dates to a 1983 flight, when pox-like erosion occurred in a forward section of the nozzle. NASA engineers estimated that the material came within 10 seconds of burning through during liftoff.

NASA implemented several changes that seemed to remedy the problem. Then, widespread erosion occurred during a May 1987 test of an aft exit cone, prompting a reexamination of earlier test results. Another cone, in 1985, was found to have eroded to a lesser degree. Both parts had been rejected for flight because of defects in the

carbon material, caused during

processing of the cone, but the severity of erosion was greater than expected, leading to the latest study.

The study group's report, dubbed a Tiger Team study because of its intensity and effort, links the erosion to excessive water and gases trapped in the finished carbon parts. It recommends changes that include curing the material longer to reduce the moisture and gases, storing parts in humidity-controlled areas and conducting to determine how the material reacts to heat and pressure.

"We can't say something bad will happen without {the changes} because we don't know," said one engineer who attended meetings on the cone. "We just don't know enough about the material and processing to say what ultimately can happen -- and that's the whole point."

NASA, which learned of some of the Tiger Team's findings in September, immediately adopted three changes in how the carbon liner is cured. The changes were implemented for the successful December test, which led some NASA officials to conclude that no other changes were necessary.

Even before the test, Larson argued that the three changes would solve the erosion problem. Adopting the others, he warned, "will result in a sufficiently different material," requiring more testing and further delays. That, he said, "is not acceptable within reasonable program constraints and is not required based on our documented experience."

In a subsequent interview, he said, "Safety is not an absolute quantity, and there is always some risk . . . . When you decide how far you want to go in a design change, you must determine how far you are going from what you know, for changes that will take long, cost more and may not be necessary," he said.

The other dissenter, G.C. Lamere, a former Thiokol engineer hired as a consultant for the study, argued in his memo that there was "insignificant data" to determine whether several of the majority's recommendations would improve the cone's performance. To approve such changes, he said,

will "unnecessarily" increase

costs.

Neither Lamere nor the authors of the majority report would comment for this report. In addition to Lamere and Larson, the members of the study group included two Thiokol employees from its Elkton, Md., plant, one from Thiokol's Wasatch, Utah, division and a second retired Thiokol engineer hired as a consultant.

Their report deals primarily with how the carbon material has functioned in the aft exit cone. Other parts of the nozzle that have more recently been questioned by NASA are not addressed.

For example, the material is also used through different processing to form the outer boot ring, which allows movement of the nozzle so it can help direct the rocket. In the December rocket motor test, the outer boot ring splintered severely. NASA has yet to determine why, and most of its concern has centered on the design of the ring,

new for that test. An alternate boot

ring design is planned for the

launch.

Engineers familiar with this latest study, however, said some scientists were troubled that the ring's design caused such stress in the carbon material, underscoring the lack of understanding of the material.

"Again, we still don't know why that boot ring failed and, again, it comes down to we still don't un- derstand what is going on with that carbon material," one engineer

said.

The carbon material is created in several steps. First, sheets of rayon broad goods are heated at temperatures in excess of 1,000 degrees Fahrenheit. The rayon decomposes, giving off hydrogen and oxygen, and a carbon cloth is formed. That cloth, black and as thick as five sheets of paper with the texture

of duck canvas, is next rolled

through a furnace that reaches about 2,700 degrees Fahrenheit, dipped into a warm resin bath and heated again.

Thiokol buys this material, which is then cut and molded into a liner for the aft exit cone, sealed in a vacuum, heated twice under pressure and fitted to the aluminum skeleton of the cone.

The report recommends that Thiokol change what it does to make the part ready for NASA, including experimenting with different heating processes, adding a third heating cycle to the finished part before it is attached to the cone, painting the parts to provide a shield against moisture and storing the cone in a humid-controlled environment.

Mitchell said many of the recommendations are "in the direction of good," but he is not persuaded that they are necessary for flight.

"Some of them wouldn't hurt, but why depart from what we already know? These things are not straightforward science. And once you get something working good, the last thing you want to do is waltz in and try something different," he said.