Analyzing unacceptable seal performance

Use this step-by-step process for seals and other equipment

Adhering to a disciplined approach to sealing system design and supporting the selection with an ongoing action plan to improve the reliability of sealed equipment dramatically improves mechanical seal performance. For many years, acceptable seal performance was defined as minimal or no visible leakage. During the last decade, these definitions changed immensely while government and industry have worked to clean up the environment. Conforming to new emission regulations is a reality that places greater attention on seal performance and the seal selection process. An action plan, when implemented in conjunction with the application of new standards and guidelines, improves seal life while reducing emissions

Many users ask, "How do I determine if a mechanical seal is performing acceptably, and what actions should I take if it is not living up to expectations?" The answer involves a disciplined, problem-solving approach and, when necessary, thoroughly reviewing the seal selection process. This action plan has eight steps:

• define acceptable seal performance,
• troubleshoot the equipment in the field,
• review the process and sealing system data,
• select the sealing system,
• investigate operational history,
• perform seal analysis,
• perform root cause analysis, and
• implement solutions and
• monitor results.

Define acceptable seal performance
Acceptable seal performance means conformance with published standards. Federal, state, and local governments regulate maximum allowable emission levels for a number of fluids. Users also create their own definition of acceptable performance for fluids not covered by these regulations, or when plant safety requires more stringent requirements.

Acceptability varies widely. Leakage from a clean water pump is usually not considered unacceptable until it is visible and defined in terms of measurable volume per unit of time. On the other hand, leakage of a volatile, hazardous air pollutant from a piece of rotating equipment is probably not visible and must be held to low levels.

The Society of Tribologists and Lubrication Engineers was the first organization to address the selection process of mechanical seals.

Troubleshoot the equipment in the field
When a seal is not meeting expectations, troubleshooting the equipment is the first step in improving the performance. It takes the straightforward, common-sense approach. Focus on corrections that can be made on-site without shutting down the equipment. Start with visual observations of the seal and the seal support system. Determine the source of the leakage and, if possible, correct it without removing the equipment from service. Typical corrective actions include turning on a flush line, adjusting a quench, tightening gland bolts, and checking pipe or tubing connections. Operational corrections include opening a suction valve to stop cavitation or adjusting a discharge valve to improve efficiency and reduce internal recirculation.

Review current process and sealing system data
If the performance cannot be corrected by actions in the field, then take the equipment out of service. Before analyzing the equipment and sealing system, however, review the current process and sealing system data as well as the repair history for the equipment. Ideally, you have an extensive database of information for each piece of equipment. The database should provide information on the equipment, process fluid, operating conditions, sealing system, and repair history. Entering the data into this database establishes a baseline from which changes can be made and progress monitored.

The review may reveal that the conditions changed to the point that the current sealing system cannot function reliably. In this case, select a new sealing system.

Select the sealing system
If you determine that the sealing system needs to be changed--at this stage or after finding the root cause of unacceptable performance-then follow established guidelines in selecting the new system. Many users establish seal selection standards based on plant experience. In addition, seal manufacturers and users worked together to establish guidelines and standards that replace or supplement a user's standards.

The Society of Tribologists and Lubrication Engineers was the first organization to address the selection process of mechanical seals. They published an "Application Guide to Control Emissions" in their Special Publication SP-30 in 1990. Many users quickly adopted this guide for selecting single, tandem, or double seals on the basis of specific gravity and permissible emission levels. SP-30 was revised in April of 1994 based on the CMA/STLE Mass Emission Study.

The revised document features an updated "Mechanical Seal Application Guide" and gives insight into the use of contacting, non-contacting, and dry-running seal technology. Since 1994, it has been adopted more widely by the user community.

The American Petroleum Institute published the world's first mechanical seal standard. Its intent is to identify the requirements of a sealing system that will run at least three years and meets emission regulations. It features a selection process for most refinery processes
and some chemical and general industry applications. As this standard is revised, most likely it will address many more applications.

Investigate operational history
Modern control rooms gather a wealth of operational data. Find out if deviations from normal operations could have played a part in causing the failure. Items to check when investigating a failure include flows, temperatures, tank levels, barrier pot pressures or levels, and motor current.Analyze trends in these variables.

This data, however, does not replace the information gained by speaking to the people responsible for operating the equipment. The operators will not only be able to discuss any operational deviations, but they will also know why the operation of the equipment deviated. For you to access this valuable information, the operational personnel must want to help solve the problem. Make them feel that they are a part of the solution rather than a part of the problem. This is not always easy. You may need to overcome years of finger-pointing among the operational, maintenance, and engineering groups.

Perform seal analysis
Disassemble and carefully clean the mechanical seal for a piece-by-piece inspection. Categorize each observation as being chemical, mechanical, or thermal in nature.

Chemical damage is usually in the form of general corrosion, pitting, corrosion-erosion (including fretting), or crevice corrosion under dynamic O-rings. Also included is chemical attack of O-rings and the leaching of the binder or secondary phase from the seal faces.

Mechanical problems develop from excessive hydraulic pressure which cause face deflection, secondary seal extrusion, or erosion. Other mechanical problems come from improper installation, uneven torque on gland bolts, or misalignment.

Thermal damage to seal faces manifests itself as heat checking, carbon blistering, pullout, setup deposits, polymerization between the faces, or phonographing, the latter of shows up itself as a series of concentric grooves. Coking, O-ring failures, and loss of spring load for bellows or springs are also considered as thermal failures.

Grouping observations in this manner permits making a simple statistical analysis that identifies the primary types of problems seen in your plant. Record all observations each time a seal is evaluated if you want the database to work effectively

Perform root cause analysis
Root cause analysis assigns the ultimate responsibility for unacceptable performance. In other words, deduce the original cause of failure on the basis of experience, troubleshooting, data gathering, and failure analysis.

For statistical purposes, organize the findings into nine primary causes, then by secondary and tertiary causes. As an example, the first classification may be "process" with a secondary cause being "poor lubrication" and the tertiary cause being "flush flow blocked or reduced." Post to the database each of the primary causes along with their secondary and tertiary causes. Analyzing these root causes and the success of corrective efforts allows making recommendations with a high degree of probability that the problem will be fully identified and corrected.

Start monitoring upon commissioning the equipment and continue under watchful eyes for the first 24 hours--the infant mortality period.

As with any statistical analysis, good data is the key to success. Good data is not always easy to obtain since the original cause of poor performance is sometimes contradictory to one's instincts. For instance, it is common to blame any dry-running pump situation on operator error. While this may sometimes be the case, dry-running can also come from poor system design, inadequate net positive suction head during startup, or it might simply be more cost-effective to allow a pump to cavitate periodically. In these cases, improve the system design or change the sealing system to make it more tolerant of actual conditions.

Implement solutions and monitor results
The first step to implementation is documentation. Federal, state, or local government regulations may require documentation of equipment design or process changes. Implementing a Management of Change process normally satisfies these regulations. At a minimum, this process requires stating the reason for a change along with necessary drawings, installation and operational procedures, updated files, and training. Once the change is approved, the person responsible for implementing it should monitor each step to ensure the process runs smoothly.

Start monitoring upon commissioning the equipment and continue under watchful eyes for the first 24 hours--the infant mortality period. Installation or operational errors are frequently the cause of failures that occur during the first 24 hours. After the infant mortality period, continue monitoring the reliability to be to ensure the implemented change produced the desired result.

Analysis of data in the database
Use the repair database to look at the repair history for a particular piece of equipment, to develop a bad actor list, or to determine the most common root causes of repair.

When you pull a piece of equipment, look at the records for past repairs. Doing so brings the analyst up to speed on problems that have been seen before on this piece of equipment, as well as on solutions that have been implemented to improve reliability.

Formulate a bad actor list by analyzing the repair data. Generate a rudimentary bad actor list by finding the pieces of equipment that have been repaired the most frequently during the chosen time frame. Include in this evaluation other variables, such as repair costs and the criticality of the equipment. In any case, generate a list of equipment causing the most problems so time and effort can be focused on them.

The repair data also determines the most common causes of repair. To do this, break down the root causes of unacceptable performance into categories for analysis. The results of such an analysis are shown below.

Analysis of root cause data
Root cause data from five chemical plants and three refineries was analyzed to determine the most common causes of repair. For statistical purposes, the findings were organized into nine primary causes, then further defined by secondary and tertiary causes. The nine primary causes are as follows:

• process,
• maintenance,
• equipment or piping,
• scheduled maintenance and the equipment did not fail,
• seal design or selection,
• seal quality,
• acceptable life-normal wear,
• other that does not fit any of the defined categories,
• cannot be determined.

Modern control rooms gather a wealth of operational data. Find out if deviations from normal operations could have played a part in causing the failure.

To break down the analysis further, the secondary causes shown below
were analyzed for the "process"
primary cause:

• operational upset,
• poor seal face lubrication,
• product setup,
• other.

• air entrained in product,
• cooling water flow blocked or reduced,
• flush flow blocked or reduced,
• heat tracing not working,
• shutdown or startup,
• barrier fluid level or pressure,
• low flow recirculation,
• low suction pressure,
• other,
• quench blocked,
• recirculation flow blocked or reduced, or
• suction screen blocked.

For statistical purposes, organize the findings into nine primary causes, then by secondary and tertiary causes.

Any cause for which a valve was closed or partially closed was put into a category called "valve closed or reduced." This includes the causes "cooling water flow blocked or reduced," "flush flow blocked or reduced," "quench blocked," and "recirculation flow blocked or reduced." The cause "heat tracing not working," which only accounted for three repairs, was also included in this group.

The tertiary causes "low suction pressure," "air entrained in product," and "suction screen blocked," were put in a category called "NPSH available." "low suction pressure" accounted for 122 of the 147 repairs in this category.

The results of the analysis are shown in Figure 5. The "valve closed or reduced" and "barrier fluid pressure or level" tertiary categories can generally be eliminated by a trained, diligent operator. The "shutdown or startup," "NPSH available," and "running back on curve" categories, on the other hand, can sometimes be more difficult to prevent.

Each of the other primary and secondary causes can be broken down similarly. Analyzing the causes of repair leads to an action plan that can be formulated to improve reliability on a plant-wide basis. A plant with a significant number of repairs attributed to "seal design or selection" should consider a seal standardization program to upgrade designs using the latest technology. A plant with a high percentage of process related repairs, on the other hand, should look into operator training programs as well as a plan to encourage operators to "take ownership" of the equipment.

Improvements achieved
To quantify the rewards achieved when following this action plan, typical improvements for a chemical plant and an oil refinery are shown in Figures 6 and 7. These improvements are shown as the percent increase in reliability to remove any controversy regarding how
mean time between repair should be calculated.

This study used mean-time-between- seal-usage as the reliability measuring stick. Every case in which a seal was used on a repair after the equipment was installed in the field and turned over to operations was counted as a repair or a seal usage. For example, if the equipment was taken out of service for poor performance or high vibration, this was counted as a usage even in cases where the seal never leaked. It would also be counted if the same seal was reinstalled during the repair. On the other hand, a coupling needing realignment was not counted.

Conclusion
A focused, disciplined approach is the key to improving equipment reliability. Following this eight-step plan lets you implement a strategy to improve seal performance. As shown in Figures 6 and 7, considerable improvements are possible. In fact, the plants in this study experienced a steady overall increase in reliability. In time, ongoing data collection can be used in other studies. For instance, one can analyze the reliability improvements achieved by converting from a component seal to a cartridge seal. This could also be done for any other possible solutions. For example, this data could be used in the future to answer the
long-standing question of whether a single seal is more reliable than a dual
sealing system.

 
Figure 1: Breakdown of root causes for all plants
 

Figure 2: Breakdown of root causes for oil refineries
 

Figure 3: Breakdown of root causes for chemical plants
 

Figure 4: Breakdown of process root causes for all plants
 

Figure 5: oor seal face lubrication breakdown for all plants
 

Figure 6: Reliability improvements at a U.S. chemical plant
 

Figure 7: Reliability improvements at a U.S. oil refinery