Protecting People and the EnvironmentUNITED STATES NUCLEAR REGULATORY COMMISSION
ACCESSION #: 9511270151
Investigation and Analysis
of
Suspect Fasteners
Event 29257
Final Report
Prepared by
Cardinal Industrial Products
Division of B&G Manufacturing Co., Inc.
November, 1995
INTRODUCTION
On July 10, 1995, B&G Manufacturing Co., Inc. ("B&G") purchased the name
and certain other assets from Cardinal Industrial Products, L.P. ("CIP-
LP") a limited partnership formed and operated in Nevada servicing the
nuclear power industry. B&G acquired these assets to begin its own
nuclear fastener business as a new division of B&G ("B&G-Cardinal").
Various owners have operated a nuclear fastener business under the
"Cardinal" name at West Oquendo Road in Las Vegas prior to B&G's
acquisition of assets from CIP-LP on July 10, 1995. In this report the
term "Cardinal facility" is used to describe the general operations which
have continued at that site independently of ownership. The reader is
advised to be aware of the nature and sequence of the ownership as this
report is reviewed.
Shortly after July 10, 1995, Duquesne Light notified B&G-Cardinal about
nonconforming Grade B7 hex capscrews, which had been processed and sold
to them by CIP-LP. Several of the suspect fasteners were tested by B&G-
Cardinal and an independent testing laboratory and found to deviate from
SA 193, Grade B7 mechanical requirements. Although these fasteners had
been processed, sold, and shipped by CIP-LP, B&G sent notifications of
the nonconformance to the NRC and to other purchasers of product from the
same lot. As a result of the notifications, two other utility companies
reported to B&G-Cardinal nonconforming fasteners from the same lot and
one additional lot.
B&G initiated a comprehensive investigation to ascertain the scope and
cause of the problem, to determine corrective actions, to keep its new
customers fully informed and, as a courtesy, to be able to instruct them
to evaluate the condition in light of 10CFR Part 21 paragraph
21.21(a)(1)(ii) and (b)(1).
Metallographic analysis of the nonconforming fasteners indicated improper
heat treatment, which prompted B&G-Cardinal to begin testing similar lots
of material from inventory that were manufactured using the same heat
treatment process. This testing revealed additional nonconforming
product and thereby indicated the problem was not isolated to only one
lot but was related to the process.
Investigation and testing traced the problem to the heat treating furnace
at the Cardinal facility, and substandard fasteners most likely can be
isolated to eight lots of material processed since 1989. (The
investigation could not address product processed prior to 1989 because
records were unavailable.)
Final Report 11/95 Page 1
This report describes the problem, the investigation strategy, testing
and analysis, and other actions taken by B&G-Cardinal that revealed the
source of the defects to be processing deficiencies in heat treating
certain material at the Cardinal facility.
THE PROBLEM
Certain lots of hex capscews that were processed and shipped per ASME SA-
193, Grade B7 out of the Cardinal facility by CIP-LP were found to
contain defective fasteners. The problem was initially discovered by
Duquesne Light's Beaver Valley facility, which had sent several 3/8"
diameter capscrews to an outside laboratory for random verification
testing. The results of these tests indicated some fasteners from the
lot were not in compliance with the mechanical requirements of the
specification. Duquesne Light reported this failure to B&G-Cardinal.
Independent test reports from the Beaver Valley facility (and later from
PG&E Diablo Canyon and Washington Public Power Supply System) suggested
the fasteners in question had not been properly heat treated.
The suspect lot of ASME SA-193 Grade B7 capscrews was manufactured at the
Cardinal facility from AISI 4140 medium carbon alloy steel. The
manufacturing process utilized by CIP-LP was to cold form capscrews from
spheroidized annealed cold heading wire. The fasteners were then heat
treated by a process of either normalizing, quenching, and tempering, or
simply by quenching and tempering. Depending on the lot sizes, the heat
treatment was either performed at the Cardinal facility or by an approved
outside vendor.
IMMEDIATE ACTION
To verify the testing results obtained by Beaver Valley, B&G-Cardinal
performed tensile and hardness tests on capscrews from the same lot.
These tests were performed at the Cardinal facility's laboratory under
the direction of B&G-Cardinal's Quality Assurance Department. The
mechanical properties of several of the capscrews tested were found to
deviate from the minimum requirements of Grade B7. This lot of material
was immediately removed from stock, marked as nonconforming, and isolated
in the nonconformance room.
Samples of both conforming and nonconforming fasteners were sent to an
outside laboratory for metallographic examination. The examination of
the conforming fasteners indicated the fasteners had the tempered
martensite structure expected from proper heat treatment. The
nonconforming fasteners, however, had a spheroidized structure that
indicated the heat-treated fasteners had not reached the temperature
required for the martensite structure to form.
Final Report 11/95 Page 2
Because the nonconforming lot was heat treated at the Cardinal facility,
B&G-Cardinal decided to concentrate its investigation on the heat
treating equipment and process there.
THE INVESTIGATION
Early indications suggested that the nonconforming fasteners had never
been heat treated. The lot in question weighed over 250 pounds and would
have been placed in several containers for ease of transport through the
Cardinal facility, so it was plausible that a lone container of fasteners
had bypassed heat treatment and was later intermingled with the rest of
lot, which had been heat treated. B&G-Cardinal then sampled a large
number of the fasteners still in inventory from the suspect lot and
found, however, substandard fasteners with visible surface scale, which
indicated that all the fasteners, including suspect ones, had been heat
treated.
Because the fasteners definitely had been heated in the furnace, the
focus of the investigation turned to the heat treatment process and
equipment at the Cardinal facility. Recognizing that a problem with the
heat treating facility may have affected other lots of material, the
Quality Assurance Department began sampling other lots from inventory
that had been heat treated through the furnace at the Cardinal facility.
During the investigation, B&G-Cardinal sent courtesy notifications to the
NRC and to companies which had purchased capscrews from the questionable
lot. Within a few days of these notifications, PG&E Diablo Canyon
informed B&G-Cardinal of a substandard, 3/8" diameter capscrew that had
come from a second lot. In addition, the inventory sampling process by
B&G-Cardinal revealed a third and fourth lot that contained substandard
fasteners of 3/8" and 5/8" diameters, respectively. Courtesy
notifications were also sent regarding these lots.
It became apparent that the four lots found to have substandard fasteners
all entered the furnace in "large" charges held at a temperature for
"short" periods (relative to other charges performed at the Cardinal
facility). The period in question was the duration parts were held in
the furnace ("soaked") at the specified temperature for austenitizing
just prior to lowering the basket of parts into a quench tank. The
soaking time for A/SA 193, Grade B7 product must be long enough to allow
all parts to achieve an austenitic microstructure prior to quenching them
rapidly in a liquid medium at a controlled temperature of 125 degrees F.
The quenching provides a rapid temperature drop that results in the
formation of a primarily martensitic structure. After quenching, parts
should register hardness values of approximately 50 on the Rockwell C
scale and tensile strengths between 225,000 and 300,000 psi. The parts
then undergo a tempering cycle that entails heating them to a
Final Report 11/95 Page 3
1,100 degrees F minimum followed by a slow cooling process. Tempering
reduces the tensile strength and develops ductility as required (for A/SA
193, Grade B7, a minimum tensile strength of 125,000 psi with ductility
exhibiting 16% minimum elongation and 50% minimum reduction of area).
Two primary problems can occur with the heat treating cycle. First, the
charge may not be allowed to "soak" in the furnace long enough for all
the fasteners to reach the temperature required to achieve an austenistic
microstructure. Second, the parts may not cool rapidly enough if the
liquid quenching medium is not circulating sufficiently enough to provide
adequate heat transfer in the required amount of time. The lots in
question were processed in heat charges of relatively heavy weight, which
could be a potential factor in either problem. If the charge size was
too large, some or all the parts may not have reached the required
temperature in the given time period; or the temperature of the quenching
liquid may have become too high to accomplish the required rapid cooling
by the time the liquid reached the parts at the center of the charge upon
being lowered into the quenching tank.
Heat treatment logs for charges performed at the Cardinal facility dating
back to 1989 were included in the records obtained by B&G-Cardinal from
CIP-LP. B&G-Cardinal examined these logs, which showed since 1989 a
total of 333 charges were heat treated involving AISI 4140 medium carbon
alloy steel. The data for each of these charges was entered into a
spreadsheet and then sorted in descending order by weight and time
(pounds per hour) with the expectation that the worst case charges
(heavier charges or shorter soak times) would appear first (See Addendum
1). The charge with the previously identified 5/8" lot with known
substandard fasteners was at the top of the list, and the other three
3/8" lots were all within the first 13% of charges listed. The pattern
of sorted data was as anticipated and indicated that sound metallurgical
principles were taking investigation in the proper direction.
Most of the product identified in the heat treatment log had been sold
and shipped prior to B&G's purchasing assets from CIP-LP, therefore lots
generally were not available from inventory for testing. Thus, the
strategy of investigation was to recreate the heat treating conditions
that would produce known defects. Test charges were planned using times,
temperatures, and weights identical to the heat charges known to have
produced defective product. Other test charges were devised with the
purpose of determining what combinations of weight and soak times would
result in defective product. From the heat treatment log, charges with
extreme weights relative to soaking times were identified for each
diameter of Grade B7 capscrews, and product was either manufactured or
taken from inventory to be used for the test charges.
Final Report 11/95 Page 4
Over a two week period, 22 test charges were heat treated. After each
charge, test coupons were removed from specific locations within the heat
treatment basket and were tested for hardness (See Addendum 2). From the
results of the early test charges, the following hypotheses were formed:
1. Some of the fasteners in the basket were not reaching the
required temperature in the allotted time. This conclusion was
based upon the location of the substandard fasteners found in
failed test charges. This hypothesis is supported by two
separate metallographic examinations of substandard fasteners.
2. The time required for all fasteners in the charge to reach
temperature is dependent upon charge density as well as overall
weight. Charge density is a function of the fastener's
dimensions where 3/8" diameter fasteners will pack more densely
than 5/8" diameter fasteners.
In addition to performing tests to verify the integrity of past charges,
other charges were specifically performed to verify the above hypotheses.
Because the problem appeared to relate in general to heat transfer and
not necessarily to any specific material, the investigation was broadened
to include all heat treated materials. The heat treatment logs were
again reviewed, but without regard for material or heat treatment type
but rather simply for what appeared to be the critical indicators of
weight, time at temperature, and charge density. No other charges were
found to have critical factors in the range which coincided with test
charge failures.
During the investigation, the possibility of a problem with the quenching
phase of the heat treating cycle was dismissed based upon metallographic
analysis. The metallography comparisons of acceptable and substandard
fasteners taken from the same lot showed two clearly different but
identifiable microstructures. The photomicrograph of the acceptable
fastener showed the proper tempered martensite microstructure with some
retained austenite. The microstructure of the substandard fastener was
spheroidized carbides in a ferrite matrix, which results only when alloy
steel does not reach the austenitizing temperature. Either
microstructure was dependent upon proper quenching. In other words,
inadequate quenching in either case would have produce yet another
identifiable microstructure.
Final Report 11/95 Page 5
INVESTIGATION FINDINGS
In the process of performing of 22 different quench-cycle test charges,
B&G-Cardinal was able to recreate heat treatment conditions and
substandard fasteners corresponding to the four inventory lots known to
have contained substandard fasteners. Because these test charges
produced the expected results and thereby substantiated the hypotheses,
B&G-Cardinal ceased investigating other causes. In addition to the test
charges that recreated known failures, other test charges produced
substandard fasteners that paralleled four additional lots of material
from the heat treatment log. While there is no conclusive evidence that
corresponding actual production charges contained substandard fasteners,
it was likely they did; therefore, B&G decided to send courtesy
notifications to the NRC and known customers for these lots, too. The
other 14 test charges done under the remaining worst-case conditions
produced acceptable results and therefore strongly suggests that
substandard fasteners can be isolated to the eight lots of inventory for
which simulated heat-treatment charges produced failures.
The test charges containing substandard fasteners were evaluated by
hardness testing of forty coupons taken from specific locations from
throughout the charge. The hardness readings were documented on forms
which detail the location of each coupon (See Addendum 2). Based upon
the map of the hardness readings, it is apparent that the bottom layer of
fasteners near the center of the basket was the last to reach temperature
in larger lots. The primary heat transfer mode for the furnace is
radiation. The secondary heat transfer mode is conduction. In large
charges of densely packed fasteners, conduction becomes much more
important. Because of the configuration of the furnace at the Cardinal
facility, the bottom center portion of the basket did not receive
significant heat through radiation. Fasteners at the bottom layer
received their heat through conduction from the upper layers of
fasteners. In the charges containing substandard fasteners, the time
required for the heat to reach the bottom layer of fasteners was longer
than the fasteners remained in the furnace.
This problem was not discovered earlier because the test coupons that had
been used by the Cardinal facility for tensile and hardness testing were
typically placed at the top, center spot of the charge. Had the test
coupons been placed within the area of the charge which failed to reach
temperature because of lagging heat transfer, the defects could have been
detected during the normal testing performed on heat charges.
The furnace at the Cardinal facility takes approximately two hours to
reach the appropriate austenitizing temperature for A/SA 193, Grade B7.
After this temperature is reached, the fasteners are allowed to "soak" at
temperature for at least one hour. The soak time is intended to provide
even distribution of the heat. The furnace's thermocouple used to
Final Report 11/95 Page 6
register its temperature is positioned in the upper region of the furnace
(See Addendum 3). The results of the test charges demonstrated that not
all the contents of a charge reach the indicated temperature at the same
time. The time required for the bottom layer of fasteners to reach
temperature was affected by the weight of the charge and the charge
density.
One of the test charges that failed (#21) was 331 pounds of 3/8" x 1"
long capscrews. This charge contained nearly 8,000 pieces and was only 2
1/2 inches deep in the basket (the basket is 37 inches long and 25 inches
wide). Another charge of the equivalent weight (#5) passed but contained
5/8" x 2 1/2" and 5/8" x 3" long capscrews. This charge contained just
over 1,100 fasteners and was 4 1/2 inches deep. How densely fasteners
were packed in the charge affected the amount of heat received through
radiation as compared to conduction. In general, the smaller the
diameter of the fasteners, the more the charge acts as a solid mass. In
both of the above listed examples the charge was held at temperature for
one hour, but the charge that failed had 3/8" diameter fasteners that
packed more densely than the charge that passed with the larger, 5/8"
diameter fasteners.
In another example, the diameter of the test charge was held constant and
packing density was changed. 300 pounds of 3/8" x 1" long capscrews were
held at temperature for one hour and contained no bad fasteners (test
charge #19). Another charge (#22) contained 270 pounds of 3/8" x 4"
capscrews and was also held at temperature for one hour. The four-inch
long capscrews; of this lot were carefully lined in rows when packed into
the basket (which is a common practice to prevent shank warpage). The
tightly packed charge of four-inch long screws contained failed
fasteners.
INVESTIGATION CONCLUSIONS
The conclusion of this investigation is that the substandard fasteners
resulted from procedural error involving the time that heat treatment
charges were held at temperature for the given conditions, specifically
the charge's weight and density as determined by the diameter of the
products and method of packing. According to the records left behind by
CIP-LP, the standard practice of CIP-LP for heat treating operations
referred to Military Standard MIL-H-6875. However, the CIP-LP's standard
practice made no reference to factoring the overall weight or packing
method when determining the correct soaking time for a particular charge
in the furnace at the Cardinal facility.
The heat treatment equipment located at Cardinal facility does have
limitations. In general, the furnace has heating elements on four sides
and on the ceiling (See Addendum 3). This would not be a problem except
that the single thermocouple is located in the ceiling of the furnace.
The temperature indicated on the display panel and recorded on the strip
chart
Final Report 11/95 Page 7
may not reflect the temperature throughout the furnace and the charge.
This limitation, however, could have been compensated for procedurally by
allowing charges to remain at temperature for longer periods of time.
B&G-Cardinal has been asked what the percentage of fasteners is
substandard in the suspect, and what is the strength of the substandard
fasteners. It should be assumed that the heat treatment of any of the
suspect lots was insufficient to achieve the mechanical properties of
A/SE 193, Grade B7, and that the mechanical properties of the fasteners
from those lots are equivalent to annealed AISI 4140 material. This
means the tensile strength is between 60,000 and 70,000 psi. Estimating
the percentage of fasteners that are substandard in a given lot is much
more difficult. Basing estimates upon heavy charges having five or six
layers of capscrews with as many as fifty percent of the bottom layer
being substandard suggests that the lot would contain 10 percent
defective fasteners. However, one cannot assume that substandard
fasteners are distributed evenly throughout a total lot; defective
fasteners, for instance, could have been concentrated into one container
when packed for warehousing or shipping. Therefore, B&G-Cardinal cannot
conclusively say what percentage of fasteners purchased by any particular
customer was substandard.
There are two other factors in connection with the heat treating
equipment at the Cardinal facility which would tend to isolate suspect
charges to the types of product and material investigated with the 22
test charges.
The first is the quenching medium, which is a polymer solution. The
particular design of the quench tank located at the Cardinal facility
precludes the use of oil because it would be a fire hazard. The polymer
solution is a suitable alternate for oil for many materials such as A/SA
193 Grade B7. But a polymer solution cannot be used in the heat
treatment of very high strength materials such as A490 structural bolts,
A574 socket screws, F912 set screws, or A354 Grade BD (SAE J995 Grade 8)
because these specifications require an oil quench. A polymer solution
is also not suitable for quenching products made from medium carbon
steels, such as A325 structural bolts, A449 Type 1 (SAE J995 Grade 5)
bolts, A194 Grade 2H nuts, A563 Grade C, D, or DH nuts, and F436 flat
washers. Consequently, there seems no reason to believe that any of
these types of products would have undergone heat treatment at the
Cardinal facility.
Secondly, rods and studs were generally not heat treated at the Cardinal
facility because they were produced from bars which already met
applicable specifications and did not require additional heat treatment.
In addition, the heat treatment basket could not physically accommodate
any studs or rods longer than 36 inches.
Final Report 11/95 Page 8
B&G-Cardinal believes that the problem of substandard fasteners
discovered initially by Duquesne Light does not extend beyond the eight
lots of Grade B7 capscrews already identified through testing charges
based upon critical factors of weight, density, and soaking times; but
B&G-Cardinal cannot guarantee this conclusion. Customers should evaluate
replacing any fasteners installed from these eight lots, and should
consider testing other lots if deemed necessary for further assurance.
SUMMARY
Shortly after receiving notification from Duquesne Light that capscrews
sold to them by CIP-LP did not meet the specification requirements, B&G-
Cardinal began an investigation into the cause and scope of the problem.
During this investigation, a detailed review of records was performed as
well as a duplication of past heat treatment charges. During the
investigation, B&G-Cardinal recreated the heat treating conditions of
suspect lots and through test charges successfully duplicated the
failures in four lots of material known to contain substandard fasteners;
additional test charges produced four more lots with substandard
fasteners, which may indicate that corresponding lots in the field may
also have substandard fasteners.
The cause of the problem was determined to be procedural error. As a
courtesy, B&G-Cardinal has notified the NRC and the companies that
purchased capscrews from lots containing suspect material.
B&G-Cardinal has discontinued using the heat treatment equipment at the
Cardinal facility until such time that satisfactory modifications are
made to both the hardware and procedures governing the heat treating
operation.
Corrective action to preclude recurrence include, but are not limited to:
1. Revising or adding procedures to address minimum soak times,
basket loading procedures, and placement of test specimens;
upgrading training programs for operators of the furnace.
2. Adding a second thermocouple to the furnace located at the
bottom-center of the furnace near the parts basket (where heat
transfer can lag behind other sections of the furnace,
depending on weight and density).
3. Adding a view port to the furnace so that the operator can
observe the charge during the heat treatment.
Final Report 11/95 Page 9
4. Considering overhauling or replacing the furnace if other
corrective actions are insufficient.
ADDENDUMS
1. Sorted Heat Treatment Data Showing Heavy Charges
2. Test Charge Results
3. Heat Treatment Equipment at the Cardinal Facility
Final Report 11/95 Page 10
Table "Heat Treatment Log Charge Data - Sorted by Pounds per Hour",
Addendum 1, Pages 1 thru 7, omitted.
Table "Test Charge Results", Addendum 2, Page 1, omitted.
Table "As Quenched Heat Treat Charge Analysis", Addendum 2, Pages 2 thru
20, omitted.
Heat Treatment Equipment
General: The heat treatment equipment at the Cardinal facility consists
of a basic electric furnace, a separate quench tank containing a polymer
solution, as well as several instrumentation and control systems.
Furnace: The furnace has two main components, the base and the door.
The base is supported by a steel structure at a height of 4 1/2 feet
above the foundation (just lower than the top of the adjacent quench
tank). The door is supported by the same structure, and moves vertically
to allow for insertion and removal of the parts basket. The door
operates hydraulically with a total travel of approximately 30 inches. A
rack supported by bricks rests in the center of the base. The basket
containing the parts being heat treated is placed on the rack. Both the
rack and the basket are made from Inconel. The furnace door houses the
heating elements and the thermocouple. There are eleven rows of heating
elements; two on each side and three across the ceiling. Side views of
the furnace door and base are shown below as Figures 1 and 2. Figure 1
is the view from the quench tank. Figure 2 is the view from the side
perpendicular to the quench tank.
Figure 1 "Furnace Side View Facing Quench Tank" omitted.
Addendum 3
Page 1
Figure 2 "Furnace Side View" omitted.
Figure 3 is a side view of the furnace with the oven door in the closed
position. With the oven door closed, there is approximately 2 1/2 inches
of clearance between the top of the basket and the upper heating
elements. The thermocouple protrudes eight inches from the ceiling and
extends into the basket approximately two inches.
Figure 3 "Furnace Door Closed" omitted.
Addendum 3
Page 2
Figure 4 is a top view (and side view) of the base with the bottom of the
figure facing the quench tank. The bricks used to support the rack are
arranged in a symmetrical pattern as shown.
Figure 4 "Top View of Base" omitted.
Figure 5 is a top view with the same perspective as Figure 4. This view
shows the parts basket placed on the rack and a cutaway view of the door
showing the clearance between the basket and the side heating elements.
Figure 5 "Top View of Base with Basket" omitted.
Addendum 3
Page 3
Figure 6 is a top view of the door as viewed from the base looking up.
The right side of the drawing faced the quench tank. This view shows the
upper three rows of heating elements and the location of the
thermocouple.
Figure 6 "Top View of Door" omitted.
Quench Tank: The quench tank is located adjacent to the furnace. The
tank contains 2,390 gallons of a polymer based quenchant. The polymer is
suspended in water and precipitates out onto the hot parts when they are
lowered into the tank. This provides an insulative layer which controls
the rate of temperature drop. The polymer returns into solution as the
parts cool. The polymer is a suitable replacement for oil as a
quenchant, except for specifications which specifically state parts must
be quenched in oil. The quenchant is circulated by an agitator assembly
located at the bottom of the tank. The agitator ensures a continuous
flow of quenchant through the parts basket during the quench. The
polymer quenchant in maintained at approximately 125 degrees F for
quenching A/SA 193 Grade B7 products.
Instrumentation and Controls: The signal from the thermocouple is sent
to a control panel which automatically maintains the furnace temperature
by controlling the power to the heating elements. The temperature is
maintained within a tolerance of +/- 15 degrees F. The temperature
signal also feeds a strip chart recorder which plots temperature against
time.
Other: The oven door, basket crane, and agitators are all operated by
individual hydraulic systems.
Addendum 3
Page 4
*** END OF DOCUMENT ***
