Sie sind auf Seite 1von 11

Mechanical Seal Reliability What Realistically Can Be Achieved

By Stephen Flood, Performance Plus UK Ltd. SYNOPSIS Many times in a single day, pump users will be asked the question why a pump has failed? Invariably the answer is because the mechanical seal has failed. Subsequently, mechanical seals are regarded as the single largest contributor to pump unreliability. However, thorough root cause analysis (RCA) would dispel this particular notion. Seal failure is in most cases a symptom of a broader problem. It has been determined that in 96% of seal failures, issues not relating to the mechanical seal were the prime cause. This paper will review modes of failure relating to mechanical seals, typical application problems and equipment selection that all contribute towards pump unreliability. Underpinning the whole process is thorough and correct RCA procedures. Without understanding the true root cause, it is unlikely the problem can be remedied.

1 INTRODUCTION In order to understand what will affect the reliability of a mechanical seal, it is necessary to understand the basic design principles and the environment that a seal should operate within. Once this is understood, the issues that affect seal performance can be identified. Variations within the seal operating environment can be driven by latent or economic factors. The requirement for increased production in order to maximise the revenue generated by an asset can lead to the equipment operating outside of its original design parameters. Mechanical seals if specified, installed and operated correctly should achieve a lifetime in excess of 36 months ref. API 682 ISO2 1049. Operation of a seal or pump outside of the conditions for which it is designed will reduce the operating life of the equipment. Performance Plus have benchmarked reliability levels at in excess of 70 process plants. Using a consistent measure, the average mean time between repair (MTBR) across the sites is in the region of 42 months. At these sites the majority of pumps will have an operating life in excess of 42 months with reliability being undermined by a small population of bad actor pumps. Reliability profiling shows a typical pareto effect at all of these sites, with 80% of failures being attributed to 20% of installed pumps. Whilst there are many different causes of seal failure, this paper will concentrate on the issues that create the majority of the failures. The root cause analysis data was extracted from the Performance Plus reliability program Interface. The data covers one hundred and forty eight different process plants and over eleven thousand failures. The data has been collected by John Crane and Performance Plus personnel on reliability contracts and alliances.

2 PRINCIPLES OF MECHANICAL SEALS There are three elements to a basic mechanical seal. The primary seal element is the fluid film interface between the primary ring and the mating ring. The secondary seal element is a dynamic seal which allows for axial movement of the seal and can be either an O ring, bellows or wedge. The third sealing element is the static tertiary seal which seals the mating ring. There is also an energizing element or spring which provides a loading force on the faces when there is no pressure present in the seal chamber. These common parts are present in all mechanical seals in one form or another to maintain the fluid film present between the primary and mating ring.

Figure 1: A Typical Pusher seal

The fluid film thickness between the primary and mating ring should ideally be between 0.5 to 3 microns. It is the fluid film that provides both lubrication and cooling to the primary seal. The fluid film guarantees low friction and consequently face wear. The thickness of the lubricating film affects both the face wear and the leakage through the primary seal. The leakage across the seal being proportional to the cube of the fluid film thickness. If the film is too thin, excess heat is generated leading to increased wear and possibly premature failure. If the film is too thick, there can be visible leakage of the sealed product. The flatness of the primary ring and mating ring also affects the fluid film thickness. In general, the face flatness should be between 1 to 3 light bands convex. As convexity increases the leakage rate also increases. Face flatness can be affected by fluctuations in temperature, pressure etc. The fluid film at the primary seal must be present and constant; variances in the conditions within the seal affect the film stability and face flatness and hence the reliability of the seal.

3 WHAT AFFECTS SEAL PERFORMANCE The main factor which affects the reliability of the mechanical seals is the temperature or temperature rise across the primary seal and its effect on the sealed liquid and fluid film interface. A commonly used primary ring material is carbon-graphite. Carbon graphite is a poor conductor of heat. Thus any heat generated by fluid shear at the faces does not soak into the body of the material. This leads to uneven thermal expansion causing the face to rotate outwards. This thermal rotation increases the leakage of the seal and chipping can also occur on the inside diameter of the face due to the higher contact pressure.

Figure 2: Positive rotation of the primary ring due to temperature.

The fluid film between the primary and mating rings undergoes an increase in temperature due to the fluid friction and viscous shear present between the rotating faces. The temperature increase can be around 20 0 C depending on the service parameters. The fluid leaking across the primary seal experiences a drop in pressure from the seal chamber pressure to atmospheric pressure. For products where the vapour pressure at the new temperature is lower than atmospheric, the product will vaporise expanding rapidly and forcing the seal faces apart. This in turn increases the leakage and decreases the temperature increase due to friction. Once the temperature has dropped below the boiling point the faces will close again. This cyclic opening and closing creates vibration and the seal spits and pops. Chipping of the faces can also occur.

Figure 3: Pressure profile across the primary seal

Where does this occur: Hydrocarbon duties above 2000C The high temperature hydrocarbon duties account for a small percentage of the pump population present on a refinery (15%) but contribute a large proportion of the failures. Typically pumps on these duties will fall into the 20% of problem applications which deliver 80% of the failures on a classic pareto chart. The mechanical seal used on the majority of these problem applications are pusher seals which use elastomeric secondary seals. The product temperature is nominally above the working limit of the elastomer, however the use of sealant support systems provides a false environment. The correct operation and maintenance of the product coolers, external flush or double seal barrier systems is essential for achieving long mechanical seal life. Failure to operate or maintain correctly the seal support systems is often the primary cause of failure. Light Hydrocarbons Light hydrocarbons are a difficult sealing duty because of the low product temperature margin. This can give rise to unstable conditions at the seal faces. Where the product temperature was above the vapour temperature the fluid flashes and the faces operate in the boundary lubrication regime which leads to an increase in temperature at the seal faces. Occasionally the high temperatures at the seal faces cause instant and complete flashing of the fluid leading to the seal faces being covered in a vapour blanket. The components of the seal will then overheat and degrade.

Pre-Flashed Crude Crude oil has a wide boiling range which makes it difficult to seal. The common problems on crude duties are hang up due to coking and/or heavy wax deposits. Thus the maintenance of steam quenches is important. Steam quality, pressure and temperature have a large affect on the effectiveness of preventing build up of coke. Non Hydrocarbon Duties The proportion of non-hydrocarbon seals present in refineries is 25 percent and are not covered by the API standards. The reliability of these duties is often the lowest on the site. The technology employed is often basic compared to the hydrocarbon duties although the product sealed can be just as dangerous.

4 MECHANICAL SEAL SUPPORT SYSTEMS The mechanical seal support systems are designed to provide the seal with the required conditions to maintain a stable fluid film at the primary seal. Ideally the mechanical seal would be able to operate reliably without requiring a seal support system however this is not possible with current technology. By installing seal services, the probability of failure increases. Services add another potential variant that must be maintained within design parameters. The API plans can be summarised as shown in the table below. API Plan 01, 02 11, 12, 13, 14 21, 23, 31, 41 52, 53A-C, 54, 74 32, 62 72, 75, 76 Description Internal system for single seals Simple recirculation system for single seals Recirculation systems with auxiliary equipment for single seals External systems for dual seals External injection systems External control system for containment seals

Table 1: Summary of common API plans

Incorrect maintenance of the seal support systems remove the required conditions for a stable fluid film at the primary seal and lead to premature failure of the seal. For example a product circulation line (Plan 11) which would ideally be a plain length of tube. However due to operating reasons may include cyclone separators, valves, flow controllers, strainers and coolers. All of which require correct installation and maintenance without which the mechanical seal will fail prematurely. Valves & Flow controllers Without a means of measure the flow there is no way of knowing what should be the correct position/ setting of the valve or flow regulator. Operators tend to adjust the valves to the closed position. Strainers Installed to remove particulates in the product but with no differential pressure gauges. The strainers are not normally inspected until after the mechanical seal fails. Coolers Highly viscous or heavy products can solidify and plug seal coolers when cold and water quality issues can cause fouling in the water side which in turn reduces cooler efficiency. Installed coolers should be checked regularly by measuring inlet and outlet temperatures.

5 REAL WORLD DATA The following data is taken from the Interface database maintained by John Crane and Performance Plus. Interface provides a centralised tool for the logging of failure information across reliability contracts and agreements. The data covers one hundred and forty eight different process plants and over eleven thousand failures over a two year period. The failures have been the subject of a thorough root cause analysis procedure the results of which were used to drive reliability improvement plans. To allow trending of the data the root causes of the failures have been grouped into Operation, Maintenance, Process and Seal. The data covers a diverse range of production processes for the refining, chemical and petrochemical industries.

Table 2: Top Level Root cause of pump failures

From the data it can be seen that the majority of the root causes of failure can be attributed to maintenance practices, processes and controls. The seal is responsible directly for only 4% of the failures. This is not to say that the seal does not fail but rather that the seal failure is a symptom of a wider problem. The majority of the failures are caused by equipment maintenance 47% and operation 37%.
Level 1 Operation Maintenance Level 2 Operation Workshop Field Level 3 Operation Workshop Installation/Alignment Bearing Seal System Process Seal Proportion of Total Failures 37% 7% 5% 13% 22% 12% 4%

Process Seal

Process Seal

Table 3: Level 3 Root Cause As A Percentage Of The Whole

Maintenance The maintenance related activities which have impacted on the reliability of the mechanical seals can be divided into three areas workshop, field and preventative maintenance.
Level 1
Maintenanc

Level 2 Workshop Field

Percentage of Level 1 15% 85%

Level 3 Workshop Installation/Alignment Bearing Seal System

Percentage of Level 2 11% 33% 56%

Table 4: Maintenance Related Root Cause Breakdown

Fifteen percent of failures of the maintenance related activities are attributed to work carried out in the mechanical workshops. There are two main causes for the failures due to workshop quality control and personnel quality. Initial build problems may be due to a lack of experience due to inadequate training or unfamiliarity with the equipment due to natural wastage of experienced fitters. However the quality control procedures should be the safety net which would prevent poor workmanship leaving

the workshop. The majority of mechanical workshops have quality control procedures (QCP) but many are not implemented properly. In order to be effective, the QCP needs to be reviewed to ensure that the correct tolerances are measured and used. For example if the process temperature has increased above the design temperature of the pump the wear ring clearances need to be increased. This has been witnessed at one European refinery where the mechanical workshop had a policy of returning the equipment to OEM specification which was causing failures due to wear ring hang up. Once the cause of the failures was identified, the pump documentation was updated and the QCP adjusted accordingly. Other common causes of failure originating from the workshop include failure to carry out dynamic balancing of rotating elements, shaft concentricity, squareness and run out, all of which effect the reliability of the mechanical seal. Where component mechanical seals are still in use the incorrect installation of the seal was a common cause of seal failure. Eighty five percent of the failures due to maintenance activities had root causes which originated from the field maintenance. The field maintenance also included the preventative maintenance activities scheduled for the equipment. The numbers of maintenance failures which are due to installation and alignment are eleven percent. These failures normally occur within six months of the pump commissioning. The pressures on the maintenance team to return a pump back to service are high and consequently the installation and alignment are often rushed. Accurate alignment of the pump and driver is essential for long bearing and seal life. During installation of the pump the mechanical seal support systems must be commissioned. The cleaning of flush lines, sealant systems, filters and orifice plates must be carried out prior to the equipment being started. Fifty six percent of all maintenance related failures are due to the incorrect maintenance and commissioning of the seal support systems. The preventative maintenance required on mechanical seal support systems includes sealant top up, maintaining pressure on barrier systems The lack of effective corrective and preventative maintenance of the pump bearings can be attributed to thirty three percent of the failures in the field. The lubrication schedules in the client business system in general have been based on manufacturers recommendations and are correct if implemented. However experience shows that apart from critical and bad actor equipment which are attended to regularly, most equipment suffers from infrequent lubrication oil checks and top ups. Possible bearing oil contamination sources include ingress of moisture through damaged lip seals which can come from heavy precipitation, pump wash down or in some cases from mechanical seal steam quenches. The fitting of modern labyrinth seals should eliminate these problems. However bearing oil contamination can also come from the source of the oil. Barrels of bearing oil improperly stored can allow water and solid contaminants to enter the oil used to top up the bearing lubrication system.

Process The process has a direct affect on the performance of a mechanical seal. Variations in the process can be categorised as upset, change and variation depending on the period

of time for which the change acts and the magnitude of the deviation from specification. A process upset or incident would be defined as a major change in process conditions outside of the normal variances which could lead to unplanned shutdown. A process change would be considered to be a long term change due to a requirement for increased production or a change in feedstock. A variation in the process would be a short term deviation from the process specification primarily driven by the operators. The response to the each of these three differs. The major process upsets or incidents can be considered to be random and in terms of the affect on pump and seal reliability negligible. The resultant corrective actions from any root cause analysis would be unlikely to drive any improvements in reliability. A long term process change due to a requirement for increased production or a change in feedstock can lead to the pump and seal operating outside of the original design parameters. The majority of new seal installations will have been well selected at the time of specification of the pump and its ancillary components. It is the change in the process over time which leads to the requirement for a new seal selection or upgrade to materials. For example a change of feedstock from sweet to sour crude can lead to a change in the chemical composition of the products being pumped and an increase in corrosion due to the incorrect materials being used. The business requirement for increased production has lead to the equipment being operated on the edge of its design envelope. When pumps are operated away from the best efficiency point reliability is reduced, and Performance Plus investigations have determined that generally the most inefficient pumps are the most unreliable. Operation The operators responsible for the daily operation of the pumps, seals and seal support systems can affect the reliability of the mechanical seals. Incorrect operation of the equipment is responsible for thirty seven percent of the failures. The common causes of failure due to equipment operation include: Steam quench and steam quality, temperature and pressure. Operators turning off the steam quenches. Incorrect decommissioning. Incorrect commissioning and vapour lock. Available Net Positive Suction Head (NPSH) versus the Required NPSH. Operating away from the best efficiency point. Operating two pumps in parallel. Throttling pumps using discharge valve.

6 ROOT CAUSE ANALYSIS A thorough and systematic approach to understanding the root cause of the equipment failure needs to be applied. The focus of the root cause analysis (RCA) is to find the real cause of the problem rather than symptoms of failure. RCA attempts to identify all contributing factors and causes. Once the root cause(s) are identified the corrective and preventative action process must then be followed and the defect can then be

eliminated. The RCA process involves collating all the factors and interviewing the people involved before examining all the available information. For example people involved in a full RCA session would include but are not limited to process engineers, operators, and maintenance technicians from both the field and workshop. The possible areas for investigation might include: Consider the current operating conditions compared to design / selection conditions. Consider the equipment within the wider process. For example is the seal operating on an external flush (Plan 32 or 62). Equipment operating procedures. Process Flow and Instrumentation Diagrams (PFID). Quality Control Procedure/Plans (QCP). Failure history.

7 LIVE EXAMPLES 7.1 A European Refinery The refinery examined here is a European refinery which has been operating a reliability contract with Performance Plus for five years. The mean time between maintenance (MTBM) for the 615 pumps under contract has been increased from 37 months in the first year to 74 months in the fifth year. The improvement has been driven by dedicated resource at the customer site provided by Performance Plus. Reliability improvement plans were put into place based on the thorough root cause analysis on all the outages. Also proactive plant tours were used to identify potential problems before the equipment failed. The root cause analysis data was used to trend common causes of failure and drive training of operators on the operation of pumps, seals and seal support systems.
P.rc.ntag. Bathtub 40% 35% 30% 25% 20% 15% 10% 5% 0% <6 6 - 12 12 - 18 18 - 24 24 - 30 30 - 36 36 - 42 42 - 48 48 - 54 54 - 60 Lif. (months) Y.ar 1 Y.ar 5 >60 N/A

Figure 4: Percentage Bathtub

Over the five year period of the contract the proportion of premature failures with a life of less than 6 months has decreased, see figure 4. The distribution of the root cause of premature failure has not changed much regardless of the increase in reliability. Year 1 Maintenance 49%, Operation 34%, Process 17% Year 5 Maintenance 46%, Operation 39% Process 15%

The results of the proactive approach to reliability improvement can be seen by comparing the root cause pie charts for all failures for the first year (Figure 5) and fifth year (Figure 6) of the reliability contract. The amount of failures which can be attributed to the maintenance of the pumping equipment and the associated support systems has decreased from 37% to 25%. The effect of the increased focus on the maintenance and operation of the equipment has been the increase in the end of life failures from 27% to 35%.
Y.ar 1 RCA L. v.l 1 Y.ar 5 RCA L. v.l 1

15 % 27%

17% 35%

21%

23%

37%

25%

End of Lif. Maint.nanc. Op.ration Proc.ss

En d

o f

L if .M ai nt . na n c .

O p .r at i on

Pr o c. s s

Figure 5: Year 1 Root Cause Breakdown

Figure 6: Year 5 Root Cause Breakdown

7.2 North African Refinery A benchmarking study was carried out at an African refinery which had a population of seven hundred and ninety three pumps. The reliability of the pumps was very low at seventeen months. The all pumps pareto chart below, Figure 8, does not represent the classic 80 20 pareto rule. The reliability of the site is very low at only seventeen months, with forty five percent of the pump population failing within twelve months. This means that the overall maximum reliability of the site could not at the time of the survey exceed twenty four months mean time between maintenance. Twenty three percent of the pumps fail more than once in every twelve months. Twenty two percent of pumps fail randomly. By the implementation of a structured and proactive approach to reliability the seal and pump life could be increased to above thirty six months over five years. The rate of improvement in seal life possible is greater than that for a European refinery because of the frequency of failure of the bad actor equipment.
Bathtub Al l P ump s Pa r. t o

120 100 80

120% 100%

80%

60 40 20 0 <3 3-6 6-9 9 - 12 12 - 15 15 - 18 Lif. (months) 18 - 21 21 - 24 >24

60%

40% 20%

0% 0% 10% 20% 30% 40% 50% # of Pumps 60% 70% 80% 90% 100%

Figure 7: Bathtub

Figure 8: All Pumps Pareto

7.3 Regional Variations The achievable mechanical seal MTBM varies according to region as can be seen from the two examples provided above. The maximum achievable depends on the starting point and the local infrastructure. The rate of improvement achievable in a second world refinery is greater due to the frequency which the equipment fails at

there is more opportunity to implement improvement plans. However the maximum achievable MTBM is limited by the following: Ease of access to Original Equipment Manufacturers (OEM) etc. Supply chain issues. Local outsourcing to machine shops and suppliers of a lower quality. Workshop and operations personnel quality. Insufficient quality control procedures. Technology. Maintenance practice. Poor workshops.

8 SUMMARY The major contributor to the reliability of pumps and mechanical seals was equipment operation at thirty seven percent, see Table 3. The correct maintenance and operation of the seal system is imperative in ensuring acceptable mechanical seal life. Failure to maintain the seal support system was the root cause of twenty two percent of all the failures examined, see Table 3. By identifying the cause of failure and addressing the problems using defect elimination techniques, the reliability can be improved. The extent and rate of reliability improvement also varies dependent upon the initial start point and the geographical location. For European plants the common starting point would be around forty two months and lifetimes in the region of eighty months are achievable. In African plants where the average MTBM is twenty eight months, the starting point can be as low as seventeen months. The level of investment is lower compared to Europe and the local infrastructure to support the plant has not been present which contributes to the reliability. The percentage rate of improvement possible for these low reliability plants is greater than that in Europe; however the end point (best in class) is lower too.

Region Europe Middle East Africa Asia Pacific North America

Average MTBM 43 34 28 37 46

Best in Class MTBM 80 50 43 61 79

Table 5: Realistic Improvement Targets For Pump Reliability

The investment required to achieve higher reliability should be offset against the benefit when deciding on the most suitable plan of action. The direct cost of repair should be considered in addition to the indirect costs associated with maintenance. A cost benefit analysis can be carried out to justify the time and costs associated with reliability improvement.

10

Direct Repair Cost: Material Labour Indirect Cost: Administration cost for raising purchasing orders. Inventory capital. Inventory management. Loss of production

11

Das könnte Ihnen auch gefallen