ACOUSTIC TESTS DURING COOLDOWN FIND CRACKS IN REACTORS AT GULF COAST REFINERY

A major Gulf Coast refinery in late 1989 conducted the largest to-date acoustic emission test for structural defects in the oil industry, according to Physical Acoustics Corp.

The project employed a large-scale cooldown, or thermal gradient, testing methodology that can be used on any number of vessels within a refinery.

The effort was designed to test for structural defects in the hydrocracker and catalytic reformer units, with minimal downtime. The testing provided a global review of structures within a time period much shorter than conventional nondestructive testing.

Acoustic emission (AE) is a nondestructive testing technique that can localize defect. AE locates defects by detecting the high frequency sound they emit while the structure is undergoing stress.

Seven reactors containing over 3,000 structural welds in 30,000 ft of pipe were checked within 36 hr using a stressing technique based on controlled cooldown from operating temperature, says Acoustic Emissions Corp. This methodology was successful in identifying a number of defects within the cooldown period.

The primary benefit to the refinery was a major reduction in the time required to get the facility back on-line (36 hr of test time with AE vs. 3-4 weeks with traditional approaches).

Physical Acoustics Corp. says that follow-up testing with conventional means (usually visual or ultrasonic inspection) takes approximately 1 day. This is because in follow-up testing, only the suspected areas are tested, rather than the entire system.

TEST METHOD SELECTION

From a technical standpoint, AE and traditional techniques such as ultrasonics, X-ray, mag particle, eddy current, and penetrants are all useful for inspection purposes. Every technique has its strengths and weaknesses.

The application of AE is based on the strength of the technique, as compared to other inspection methods. Table 1 outlines these key points.

Although each method has its benefits, turnaround time is of the essence for a refinery. Ultrasonic testing of over 3,000 structural welds may be as good at finding defects as AE, but it will take significantly more time. Each inch of weld must be manually inspected. This approach can mean not only more time, but also scheduling and manpower-related delays.

ACOUSTIC EMISSION

The need for acoustic emission testing arose when the refinery's experience indicated a high level of weldrelated cracking in chromiummolybdenum (Cr-Mo) steel reformer reactors.

Extensive pipe weld cracking was also observed in high-pressure reactor feed/effluent circuits of hydrocracker units during previous downtime. These failures were the result of thermal stress on pre-existing fabrication-related weld defects, such as lack of fusion. Metallurgical analysis of weld cracks showed a high probability of additional occurrences.

There was concern about the tendency of some Cr-Mo steels to undergo creep embrittlement in the heat-affected zone of welds, which causes cracking and weld failures.

It was also possible for the shell of the hydrocracker reactor to undergo temper embrittlement.

Due to the high pressure/temperature (2,000 psi, 900 F.) nature of the process, this created the potential for a significant failure.

AE INSPECTION

The real strength of AE is that it gives the inspector a specific area to check. Location algorithms can narrow down the source of AE activity to within a few feet.

In large structure testing, an array of sensors is placed at 20-in. spacings over the surface. The sensors are attached to the surface using hot glue. They are strategically placed to form triangles and parallelograms. Proper placement of the sensors is learned through experience, according to Physical Acoustics Corp.

On insulated structures, AE has the added benefit of requiring only a 4-in. hole for access. For many conventional inspection methods, the cost of the insulation removal can be greater than the cost of the actual inspection. If the insulation material is hazardous, like asbestos, the costs are even greater.

In the case of structures over 200 F., waveguides are typically used to satisfy the sensor temperature requirements. The waveguide is mounted to the structure with magnets or clamps, while the unit is in operation. The acoustic signal propagates up the waveguide to the sensor, which is at ambient temperature.

AE TEST PLAN

Seven reactor vessels and associated piping were tested. This hydrocracker system is shut down approximately every 2 years for catalyst regeneration. At the last outage, an AE test using hydrostatic pressure of 1.5 times reactor design pressure was run on the reactor feed/effluent piping circuits.

Hydrostatic testing proved to be worthwhile in identifying defects, but required additional time and effort associated with blinding piping and general setup. An innovative approach to stressing the piping and reactors through a controlled cooldown was utilized. This technique is called thermal gradient acoustic emmission testing.

Stresses develop in the structure during cooldown because of the temperature differential, or thermal gradient, between the inner and outer surfaces. System pressure is maintained at operating level throughout the cooldown.

Because large structures sustain significant loads as a result of physical shrinkage during cooldown, major stresses can build between the inner and outer surfaces when the cooldown rate is too fast.

The rate of most cooldowns is governed by the manufacturer's recommended cooldown rate for the thickest-walled vessel. Without a proper cooldown rate, cracking can develop as a result of the buildup of internal stresses.

The cooldown rate is typically similar to that used by American Society of Mechanical Engineers (ASME) boiler and pressure vessel code, section VIII, which governs the heat treatment of pressure vessel steels. However, in determining the actual cooldown rate, consideration needs to be given to the effects of the process environment and to the complexity of the structure.

Based on a maximum 8-in. wall thickness, the cooldown rate was calculated to be 50 F./hr. This rate, and the resulting stress, can be calculated, based on the wall thickness, the temperature differential gradient between the walls, and the material's coefficient of expansion and modulus of elasticity.

A maximum 100 F. temperature differential was allowed between the cooling process gas and the outside surface.

The goal of the controlled cooldown is to keep stress levels at 1.1-1.5 times the design limit of the structure. Common industry practice is to monitor the process gas temperature, but not necessarily the metal skin temperature, during cooldown.

The cooldown is accomplished by shutting down the feed to the unit and continuing to circulate process hydrogen. The cooling rate is controlled by adjusting the hydrogen temperature with the unit heaters. At the end of the test, nitrogen is used to purge the system.

Controlling the cooldown rate was expected to reduce the chance of producing thermally induced defects.

Although it was expected that the overall cooldown would take slightly longer due to this feedback control, it was found that the cooldown went faster than past efforts. This was attributed to more precise control of metal temperatures and cooling rates through the use of multiple skin thermocouples.

TESTING

The planning and setup associated with the AE test took approximately 3 weeks. The cooldown monitoring took place over a period of 36 hr. During the test, over 600 sensors were used to monitor the vessels and piping.

A portion of these sensors is shown in the schematic of part of the hydrocracker reactor sections (Fig. 1). The area depicted is about 1,000 ft long and 100 ft wide, but represents only part of the total area tested. Each sensor is connected by a single cable to instrumentation and computers, which are located in vans on the site.

Of particular importance is the parallel processing nature of AE instrumentation. Because AE activity can come from any sensor at any given moment, it is important that the computer architecture support a high data throughput.

Physical Acoustics Corp. of Princeton, N.J., has developed its instrumentation to support up to 128 channels at one time.

The PC-AT-based operator interface provides a real-time display of AE activity to allow the operator to monitor the process against loading. All AE equipment used in this cooldown testing was manufactured by Physical Acoustics Corp.

Fig. 2 is a representation of the way a weld defect appears on a printout,

One hundred sixty thermocouples were also used to monitor the external metal skin temperatures of the structures. This allows the applied stress, in terms of thermal gradient, to be correlated with acoustic data.

Over 50 inspectors and support personnel were used in this effort. Two third-party companies actually conducted the tests. MQS Inspection Inc. of Houston had primary responsibility for the hydrocracker, and Det Norske Veritas (DNV), Houston, had responsibility for the reformer.

Overall, 15,000 man-hours were expended during the test, on all aspects of the program, including scaffolding, cable routing, waveguide installation, thermocouple preparation, and complete teardown. Testing the hydrocracker proved to be the most significant task, requiring over 10,000 man-hours.

As with any AE test, follow up with conventional techniques was critical to the project. Eighty percent of AE indications on the reformer and hydrocracker were verified as cracks in piping or reactor structures. Fourteen cracked welds, f rom 1 0 to 1 00 i n. long and 0.5 to 4 in. deep, were identified in the reformer. The hydrocracker showed 20 weld cracks ranging from 6 to 76 in. long and 0.5 to 2 in. deep.

One lesson learned, however, was that numerous support structures, such as insulation ring tack welds, also gave off AE from cracking. Although verified as actual cracks, these supports have no structural significance and caused minor time delays in verification.

In future construction designs, attachments will be made in areas away from welds to prevent the need to verify these emissions.

Another double-check of the cooldown stress methodology was made on a portion of the hydrocracker piping system. Because this area's cooldown rate was difficult to control, both a cooldown and hydrostatic test were conducted on the same structure. As expected, the results from both tests showed identical defect locations.

REFINING EXPERIENCE

According to Thomas Farraro of Citgo Petroleum Corp., Lake Charles, La., and Claudio Allevato of DNV Industrial Services Inc., Houston, in a paper presented at the National Petroleum Refiners Association maintenance conference last month, thermal gradient AE testing has many advantages over traditional inspection techniques.

One of these advantages is that AE test data are analyzed using proprietary software designed to yield a severity or failure potential rating for each source identified. Conventional test data are evaluated by comparing the defect size, orientation, geometry, and type to "acceptable" defects, as defined in generally accepted engineering codes and standards.

Most of these codes were developed to ensure the quality of fabrication of new process equipment, and were not intended to be used to evaluate defects that develop in service as the result of exposure to a particular process environment. This makes them somewhat inapplicable to cracking caused by exposure to refinery process environments.

AE testing using thermal gradient stressing sequences has been proven to be valuable in the inspection and evaluation of the structural integrity of high-temperature refinery process equipment. Fig. 3 shows typical process parameter trends during thermal gradient AE testing.

According to Farraro and Allevato, the main purpose of the stressing sequence is to assure the creation of a new stress field around potential defective areas. Under these circumstances, the generation of acoustic emission indicates the existence of defective areas that are sensitive, or were activated by the stressing sequence.

With AE testing, one can globally inspect process equipment, physically locate active or growing defects, and evaluate the failure potential of those defects under actual operating loads, all in a single testing sequence. It provides the refinery engineer/inspector with a unique source of information on the safety and reliability of the equipment being tested.

At the same time, this information can be obtained with a minimum disruption to normal operation and maintenance routines.

Thermal gradient AE testing for the evaluation of high-temperature refinery process equipment has been utilized, to a limited extent, by U.S. refineries since 1985. Initially, tests were limited to individual vessels or small sections of piping.

The first large-scale test was conducted in 1988, when the entire reactor feed/effluent circuit of a fixed bed catalytic reformer unit was tested, including the reactor vessels, feed/effluent heat exchangers, and piping. The test was successful in identifying several thermal fatigue cracks in the piping and reactor vessels.

The success of this test, along with recent advances in acoustic emission testing hardware and software technology, have contributed to the acceptance of thermal gradient AE testing as a viable and useful technique for the evaluation of refinery process equipment.

Since November 1988, over 25 large-scale thermal gradient AE tests have been conducted at refineries in North America, say Farraro and Allevato. The results of the 14 most significant of these tests are summarized in Table 2. The majority of this testing has been conducted on catalytic reformer unit reactors and equipment.

Although AE sources were identified in every test, follow up inspection using traditional inspection techniques (primarily angle beam ultrasonics) indicated significant cracking in nine of the systems tested. The AE sources found in the other five systems were attributed to defects that were evaluated by traditional nondestructive testing techniques as being noncritical to operational reliability

Individual application of AE technology depends on specific parameters such as the criticality of the structure, the cost of lost production time, the availability of equipment, and the logistics of planning AE testing.

It is currently expected that more AE testing will be planned in conjunction with future scheduled outages.

Copyright 1991 Oil & Gas Journal. All Rights Reserved.