Shell and tube exchanger failures spur reliability programs

Out of a total of 1,164 exchangers, 534 experienced some type of failure mechanism. Fig. 1 shows the distribution of all the failure mechanisms.

Fig. 2 shows the distribution of failure mechanisms in older and newer plants.

In the 30-year-old plant (Fig. 2a), most of the cooling water failures occurred on the ID of carbon steel tubes as a result of underdeposit corrosion. Of the 47 exchangers with cooling-water failures, 15 had cooling water on the shell side. Of the 18 exchangers with tube failures as a result of process corrosion, 40% were upgraded in metallurgy.

The 22-year-old plant had 140 exchanger failures out of 266 exchangers (Fig. 2b). As in the 30-year old plant, most cooling-water failures were on the ID of carbon steel tubes as a result of underdeposit corrosion. Of the 19 SCC failures, 16 failures were a result of cooling water being on the shell side.

Of the exchangers with process corrosion failures, 70% were upgraded in tube metallurgy. The mean time to failure (MTTF) for steam or condensate corrosion failure was 17 years. All four erosion failures were at the reboiler inlet.

The newest plant surveyed, one of 6 years, had 33 exchanger failures out 238 (Fig. 2c). Of the 12 SCC tube failures, 7 failures occurred on stainless steel tubes, which had cooling water on the shell side. Five of these exchanger tube bundles were upgraded to SS2205.

Admiralty brass tubes accounted for 4 of the 12 SCC failures.

A Weibull analysis of cooling water exchanger failures was performed for each production unit. The range of MTTF was found to be between 7.2 and 13.2 years.

The team identified three main areas for improvement to minimize future in-service tube failures. These areas were:

1. Tube testing
2. Retubing strategy during turnarounds
3. Design improvements.

Tube testing

Technologies for inspecting heat exchanger tubes are rapidly changing and continuously improving. Lyondell`s experience has shown a considerable variance in test results depending on both the instrument and operator.

The impact of human factors on non-destructive examination (NDE) performance is more prominent with the increased complexity and sophistication of today`s NDE techniques. Root cause analysis of some of the tube failures in the last few years identified the need for reliable technicians and technologies.

The team began studying tube-testing technologies by reviewing and evaluating the effectiveness of current NDE techniques. Different techniques are used depending on tube material, tube cleanliness, type of defects expected, and plant-inspection department preference and experience.

The preferred method for inspecting non-ferromagnetic tubes is conventional eddy current. This technique is well-established in the NDE community and is recognized by ASNT (American Society for Nondestructive Testing) and ASTM (American Society for Testing & Materials) as a standard means for inspecting non-ferromagnetic tubular products. Lyondell has used this technique since the late 1980s and has high confidence in it.

Ferromagnetic tubes, however, present a more challenging test environment. Tube scaling and variation in magnetic permeability are the main reasons for less-accurate results. Lyondell has experienced high variability in the NDE results on carbon steel tubes. Sometime the operator`s skill was questionable.

No national organization has recognized or standardized a technique for the testing of ferromagnetic tubes. Currently, four techniques are being used at Lyondell to test ferromagnetic tubes. They are:

1. Remote field eddy current testing (RFEC)
2. Partial saturation eddy current testing (PSET)
3. Magnetic flux leakage (MFL)
4. Internal rotating inspection system (IRIS).

Each technique has its own limitations. Industry studies and Lyondell`s experience have shown a considerable variance in detectability and accuracy.

The team developed a program to qualify NDE technology, hardware, and operators through performance tests. The purpose of this program was to test, rank, and use only the top tier technicians, hardware, and technologies to obtain reliable tube-test data during turnarounds.

A mock exchanger was fabricated using the most common tube sizes and materials found in Lyondell facilities. Each tube has a known set of flaws. The types and location of flaws simulate actual situations.

In addition, tubes removed from in-plant exchangers with actual flaws were used to satisfy quantitative needs and to simulate the liftoff effect that exists in dirty tubes. Random placebo tubes (tubes with no defects) were installed to further validate operator performance.

Operators regularly used at one plant conducted the preliminary tests. Fig. 3 shows the results of testing performed by a level three operator. The graphs show the actual physical defect vs. measured defect size. The scatter is highest for the remote field eddy current-testing technique and the magnetic flux leakage-testing technique. This is a result of the limitation of these techniques in detecting certain types of defects.

The current plan is to test all operators, hardware, and technologies for the next turnaround to establish a base line for qualification. This process will be fine-tuned at every turnaround.

For consistency, the qualification program will be coordinated by one individual who will execute the testing protocol, grade NDE technicians, and maintain a database of tested individuals.

Retube strategy

In the past, the turnaround scope for tube testing and retubing was developed using a narrow set of criteria and that did not include the business or safety/environmental consequences of a tube leak. The number of unplanned retubes was high.

The team realized that a formal risk-based approach is required to prioritize bundles for inspection and retubing based on both failure potential and failure consequences. A retube-analysis matrix was developed to provide a systematic and consistent approach to predict problem heat exchangers and to develop a thorough turnaround scope.

During the turnaround scope development process, Lyondell requires a multidisciplinary team from operations, maintenance, and reliability engineering review and endorse the inspection and retube priority list.

The retube-analysis matrix is based on a point system. The maximum number of points that can be obtained is 16; 0 is the lowest risk and 16 is the highest risk. The following four risk factors are used in the matrix:

1. Tube-age factor
2. Remaining life factor
3. Production-criticality factor
4. Service factor.

The tube age factor compares the current age of the tube bundle to the historic average age of the tube bundle. The remaining life factor is based on the current tube wall loss and the corrosion rate.

The production criticality factor takes into consideration the business and/or safety/environmental consequences of a tube leak. The service factor primarily covers the variables associated with reboilers, cooling water exchangers, and the mechanical design. A maximum of 4 points can be obtained for each factor.

The inspection and retube priorities are determined based on the total number of points. The following are the three risk categories:

1. Retube, 11-16 points
2. Priority inspection, 8-10 points
3. Normal inspection, 1-7 points.

Retube-category exchangers are high-risk exchangers and are planned for retube during the turnaround. Exchangers in the "priority inspection" category are scheduled for testing early in the turnaround so that if required, there is sufficient time during the turnaround for retube.

The turnaround team must evaluate whether material is required and/or shop space must be reserved for these exchangers. "normal inspection" category exchangers are not expected to have problems and are inspected after the priority exchangers.

Fig. 4 (at right) shows the retube-analysis matrix. Fig. 5 (below) compares the scope developed for a turnaround using traditional practice and using retube-analysis tool.

The retube-analysis matrix was validated and fine-tuned against past in-service tube failures (reliability hits) and past turnaround scopes. Of the exchangers that failed in service resulting in a production loss, 77% would have been retubed during the turnaround if the retube-analysis matrix had been used. The remaining 23% of the exchangers fell in the "priority inspection" category.

Of the unplanned retubes during the turnaround, 91% fell in the "retube" or "priority inspection" category. Fig. 6 shows the validation of this tool against past in-service failures and unplanned retubes during turnarounds.

Design improvements

The team identified a number of improvements to add to the company heat exchanger engineering standard to improve on-stream heat exchanger reliability. The new standard includes the following requirements:

• Duplex stainless steel tubes should be used for shell-side cooling water exchangers. Generally, the cooling water is on the tube-side.
• Fouling process services identified by the team must be on the tube side.
• Stainless steel internals, such as baffles, tie rods, and tube supports, should be used in exchangers with alloy tubes and cooling water or steam on the shell side.
• There must be a 3/4 in. minimum diameter for carbon steel tubes.
• Removable U-tube bundles are required instead of fixed or welded U-bundles.
• Two rows of solid impingement rods must be used instead of impingement plates for steam reboilers. The maximum allowable rv2 value at the entrance area was reduced to 2,400, where is the steam density and v is steam velocity.
• Strength welding of the tube-to-tubesheet joint is required in services identified by the team.
• There must be 90-10 Cu-Ni material for packaged chiller units.
• Inconel 600, Incoloy 800, or Inconel 825 expansion bellows should be used to avoid external SCC.
• Socket-welded tubesheet vents and drains must be included.
• There must be steam out design for all process exchangers.
• There are tighter flatness requirements on body flanges in higher temperature heat exchangers.
• Serrated metal gaskets (for example, Kammprofile gaskets) must be used for body flanges.

The Authors

Carlos Gamio is turnaround manager for Lyondell Chemical Co. Currently, he is managing a large turnaround in the Houston area. Previously, he held positions as reliability engineering superintendent, maintenance superintendent, and inspection supervisor with Lyondell, formerly known as ARCO Chemical Co.

Prior to joining Lyondell in 1990, Gamio spent 8 years at Exxon Co. U.S.A. He is a registered professional engineer in Texas and holds two BS degrees in engineering from Texas Tech University, Lubbock, Tex.

Walter Pinto is manager of stationary equipment engineering at Lyondell Chemical Co. His group provides worldwide technical support in the areas of stationary equipment, mechanical integrity, inspection technology, and materials engineering.

He has 13 years of industry experience in the design and fabrication of vessels, exchangers, and piping. Pinto holds an MS degree in mechanical engineering from University of Wisconsin, Milwaukee.