|Year : 2012 | Volume
| Issue : 2 | Page : 129-134
Failure Mode Analysis of Mechanical Seals
Jasbir Singh, Surjit Angra, Vinod Kumar Mittal
Department of Mechanical Engineering, National Institute of Technology, Kurukshetra, India
|Date of Web Publication||4-Aug-2012|
Department of Mechanical Engineering, National Institute of Technology, Kurukshetra
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Mechanical seals are important to prevent leakage and entry of foreign particles into the system. Therefore, failure of mechanical seals can be hazardous to the system. Understanding the various failure modes or symptoms will facilitate improvement in design, operation, and reliability of the mechanical seals. Failure modes of mechanical seals are analyzed by using the 'Digraph Modeling and Matrix Approach'. Mechanical seal failure logic diagraph (MSFLD) is prepared for failure modes of mechanical seals and is described by contributing cause events (direct and indirect) and their inter-relations. The connection and reachability matrix is used to analyze the digraph model. The mechanical seal failure connection matrix (MSFCM) is obtained from the MSFLD, and the MSFRM is derived from it, which helps in identifying the stage relationship among various cause events and their importance. This helps in taking appropriate steps to minimize the failure and improve the reliability of the mechanical seals.
Keywords: Failure mode analysis, mechanical seals, cause events, diagraph and matrix approach
|How to cite this article:|
Singh J, Angra S, Mittal VK. Failure Mode Analysis of Mechanical Seals. J Eng Technol 2012;2:129-34
| 1. Introduction|| |
Mechanical seals are used in industrial pumps, compressors, and other applications, to provide a leak proof seal between the component parts. There are many different designs for mechanical seals, to meet specific applications. A mechanical seal  is a dynamic design with spring elements mounted in the dynamic (rotating) part of the seal system, or it can also be a stationary seal with spring elements mounted in the stationary part of the seal system, to compensate for misalignment of the shaft and seal.
Mechanical face seals are used to seal a fluid at places where a rotating shaft enters an enclosure. A rotating seal is fixed to the shaft and rotates with it, whereas, a stationary seal is mounted on the housing. The secondary seals prevent leakage between the rotating shaft and the rotating seal, and also, between the housing and stationary seal, respectively. The rotating seal is flexibly mounted in order to accommodate angular misalignment and is pressed against the stationary seal by means of the fluid pressure and the spring. Primary sealing occurs at the sealing interface of both seal faces, where the rotating face slides relative to the stationary face. For proper functioning of a mechanical face seal, a fluid film is maintained between the faces. The sealed fluid may also act as a lubricant.
The failure of the mechanical seals can be understood easily by identifying the various cause events that lead to their failure. Different cause events could initiate a general or specific failure mode. The reliability of mechanical seals can be increased by identifying failure occurrence and its propagation. A few cause events that are crucial can be minimized by appropriate actions at the design or fabrication stage. Structure  is the key to understand the failure of a system / components and its propagation. Structure or topology may be physical or abstract. Physical structure implies its components / assemblies and their connections, while an abstract structure involves failure contributing events or parameters and their interconnections or interdependences. A well-established approach, that is, the digraph model, suggested in graph theory [3,4], is used to represent structure (physical or abstract). A failure mode of mechanical seals can be conveniently represented in terms of a digraph model, which consists of nodes and directed edges. The digraph model for large systems is very complicated. To analyze the digraph model, a computer is used as a tool, and the analysis provides direction for the minimization of failure modes. The connection and reachability matrix approach is used to analyze the digraph model. Sehgal et al. [5,6] have applied this approach to analyze the failure of welded joints and the rolling element. This approach is extended to the failure mode analysis of mechanical seals, in this article.
| 2. Mechanical Seal Failure Modes and Cause Events|| |
A mechanical seal may fail during its service by wear, corrosion, fracture, and so on. In a plant, leakage from a machine or machine components is reported as a failure mode or symptom. A cause event that singly or in combination with other cause events leads to various failure modes or a general failure mode, includes, design-related, process-related, and operation- and maintenance-related cause events. The list of cause events is prepared after consultation with design and practicing engineers. Some of these cause events will directly influence the failure mode of a mechanical seal, whereas, the others will be influenced by a stage relationship.
2.1 Cause event relationship
The cause events, that is, a problem of abrasive contaminants, improper seal material, problem of maintaining pressure velocity coefficient (PV value), excessive fluid pressure, improper seal design, insufficient seal lubrication, excessive seal face temperature, and failure of secondary seal [1, 7, 8], lead to seal failure directly. For example, excessive seal face temperature leads to high heat generation causing seal failure. Similarly insufficient seal lubrication leads to seal failure because inadequate lubrication causes high heat generation in the seal, which leads to seal failure. Furthermore, failure of the secondary seal by spring failure causes the seal to face away from the stationary as well as the rotating parts, leading to leakage of fluid, which is unacceptable and it is a symptom of seal failure. When abrasive contaminants get into the system, they lead to the wear of the mechanical seal, and its efficiency to operate reduces. These cause events are called direct cause events. Other cause events such as problem of misalignment, improper seal assembly, and installation and poor maintenance practices are called indirect cause events. For example, poor maintenance practices such as inadequate cooling lead to an increase in the seal face temperature, which further causes seal failure. Misalignment causes a contact between the seal faces and the contact part, like the shaft, due to which friction increases, and hence the temperature increases at the contact surfaces leading to seal failure. Some of the cause events directly lead to seal failure, whereas, others influence these cause events through various stage relationships. [Table 1] gives the contributing cause events and their relations, for a seal failure. The second column, lists 11 cause events, whereas, the third column pinpoints the cause events that directly influence seal failure. The fourth column gives the interrelations among the cause events, which are directly influenced by the corresponding cause events given in the second column. For example, cause event 4 (excessive fluid pressure) is related to cause event 3 (problem of maintaining 'PV' value), cause event 5 (improper seal design), cause event 10 (failure of secondary seal), and cause event 11 (poor maintenances practices). Out of these cause events, cause event 4 (excessive fluid pressure) directly influences cause event 3 (problem of maintaining 'PV' value), cause event 5 (improper seal design), and cause event 10 (failure of secondary seal), and it gets influenced by cause event 3 (problem of maintaining 'PV' value) and cause event 11 (poor maintenance practices). In order to represent the cause events, and the direct and indirect influence among cause events for mechanical seal failure, as shown in [Table 1], it is desirable to use appropriate modeling. Digraph modeling is one such approach and is also convenient to represent.
| 3. Digraph Modeling|| |
The cause events and their relations identified in the preceding section for seal failure are represented conveniently in terms of a digraph, for its analysis. A digraph called the mechanical seal failure logic digraph (MSFLD) is defined by considering all the possible contributing cause events and their interrelationship. Nodes in the digraph [3,9] consist of contributing cause events to the mechanical seals.
Seal failure and the edges represent their interrelations. If a cause event i is related to j, that is, it facilitates the occurrence of seal failure, then node i is connected to node j by a directed edge from i to j (i.e., e ij ) in the MSFLD. It means that cause event i will initiate the cause event j, which may lead to another event or ultimately lead to seal failure. It means that the edges in the digraph represent the causality relation among the cause events. The direct cause events for the seal failure are problem of abrasive contaminants, improper seal material, problem of maintaining 'PV' value, excessive fluid pressure, improper seal design, insufficient seal lubrication, excessive seal face temperature, and failure of secondary seal. The remaining cause events, although not influencing or connected to seal failure directly, are interrelated and related to the direct cause events by a stage relationship. Mechanical seal failure logic digraph (MSFLD) is developed by using [Table 1] and is shown in [Figure 1]. The mechanical seal failure logic digraph [Figure 1] corresponding to seal failure starts with a dummy node and ends with a seal failure, and visually represents the cause events and their causality relationship.
|Figure 1: Mechanical seal failure logic digraph. Events: 1. Dummy, 2. Problem of abrasive contaminants, 3. Improper seal material, 4. Problem of maintaining PV value, 5. Excessive fluid pressure, 6. Improper seal design, 7. Problem of misalignment, 8. Improper seal assembly and installation, 9. Insufficient seal lubrication, 10. Excessive face temperature, 11. Failure of secondary seals, 12. Improper seal maintenance practices, 13. Mechanical seal failure|
Click here to view
Mathematically a mechanical seal failure logic digraph (G MSFLD , i.e., MSFLD) is defined as:
G MSFLD =[N, E]
Where N=[N 1 N 2 ,…, N n ]
and E=[e ij ]; i,j=1 to n with i≠ j
The node [N i ] represents its failure cause event with i=2, 3, 4, n-1. The node N n represents the seal failure and N 1 is a dummy node, and edge e ij , represents the interconnection or interrelation among the cause events i and j, for the seal failure. A directed edge e ij for i=2 to n-1, and j=n represents the direct influence of its cause event on seal failure, whereas, a directed edge e ij for i=2 to n-I, with i ≠ j, represents the indirect influence of cause event i on the cause event j.
A mechanical seal failure logic digraph is structure dependent. A visual interpretation of cause relations of seal failure can be graphically represented by a digraph model. As the number of contributing cause events increases, analysis of digraph models becomes complicated. This problem is overcome by the use of the matrix approach. The digraph can be represented in a matrix form, which is conveniently handled on computers. When representing the failure logic digraph by a matrix directly, without actually drawing the digraph, one may overlook some multistage relations. Therefore, it is important to develop the digraph and then represent it by a matrix. The equivalent connection matrix of the digraph is represented for the analysis of seal failure. It is called the mechanical seal failure connection matrix (MSFCM).
| 4. Mechanical Seal Failure Connection Matrix|| |
This matrix is defined for the mechanical seal failure logic digraph (MSFLG). It represents a node-node connection of the digraph. The off diagonal entities C ij , i≠j, and j≠n, show the relation between cause event i and j. In addition, C ij =1, for i=2 to n-1, and j=n, represent the cause event I, which directly leads to seal failure. C ij =1 if i and j are connected together with an edge from i to j and C ij =0, otherwise, the diagonal entities of the matrix are zero, as the matrix shows only the connections or causality relation among the contributing events of the seal failure. For the mechanical seal failure logic digraph [Figure 1], the mechanical seal failure connection matrix (MSFCM) is written as:
Mechanical seal failure connection matrix (MSFCM) is represented by 'CM'. As there are 13 nodes in MSFLD [Figure 1], it is a 13×13 matrix. The mechanical seal failure connection matrix (MSFCM), that is, matrix (1), shows only the direct relation between the cause event and the seal failure or between two cause events, but does not indicate the indirect influence, that is, stage relationship between the cause event and seal failure. It does not, therefore, explicitly provide the sequence of cause events leading to seal failure. This information is useful in minimizing the failure at the design or operation / service stage.
In order to obtain the information, we need to convert the matrix (1) [MSFCM] into a Reachability matrix. The Reachability matrix [10-12] of a digraph contains all the stage relations between the various nodes (cause events and seal failure or between two cause events) of the digraph, which is an important characteristic of the MSFLD. In view of this, the mechanical seal failure reachability matrix (MSFRM) is defined for processing of the reachability matrix representation of the digraph that finds all the paths from every node N i to any node N j within the digraph. Various algorithms [13,14] have been developed to perform the reachability calculations. In the present study, the algorithm suggested by Busaker and Saaty  is used. The algorithm is discussed stepwise in the following section, to obtain the mechanical seal failure reachability matrix (MSFRM) corresponding to the mechanical seal failure logic digraph (MSFLD), as shown in [Figure 1], and mechanical seal failure connection matrix MSFCM in matrix (1).
| 5. Mechanical Seal Failure Reachability Matrix|| |
Mechanical seal failure reachability matrix (MSFRM) for mechanical seal failure connection matrix (MSFCM) is obtained as follows:
Step 1: Multiply the Connection matrix [CM] with itself from the right hand,
That is, CM2=[CM] × [CM]
Step 2: Make the non-zero elements of the matrix obtained in Step 1, unity, and make the diagonal elements zeroes because diagonal elements are insignificant.
Step 3: A non-zero or unity element, which appears for the first time in CM2, implies that there is a two-stage relationship between the cause event of a row and the cause event of a column, and thus, can be represented as 'n' in the final mechanical seal failure reachability matrix (MSFRM)
Mathematically, R 2 =CM2=[CM] × [CM]
Step 4: Repeat Step 2 until no more new matrices appears. A non-zero or unity element that appears for the first time in CM n implies that there is an n-stage relationship between the cause event of a row and the cause event of a column, and thus, can be represented as n in the final mechanical seal failure reachability matrix (MSFRM).
Step 5: Write the final matrix called the mechanical seal failure reachability matrix (MSFRM) by including not only the various non-zero / unity elements obtained during various stages and represented as n (corresponding to an n-stage relation), but also the unity elements of the matrix CM, that is, matrix (1).
From mechanical seal failure connection matrix, that is, matrix (1), the mechanical seal failure reachability matrix (MSFRM) is obtained after seven multiplications, and is given as:
The mechanical seal failure reachability matrix (MSFRM), that is, matrix 2, shows a one-, two-, three-, and four-stage relationship between two different cause events or cause event or seal failure, where a one-stage relationship shows a direct relation, while two-, three-, and four-stage relationships show an indirect relationship, as shown in matrix 2. These stage relationships for a cause event are obtained from the reachability matrix, by looking at the column corresponding to that cause event. For example consider the cause event C 10 (excessive seal face temperature); the event that is connected to cause event C 10 is most immediate, that is, the one-stage relationship, which is cause event C 3 (improper seal material), cause event C 9 (insufficient seal lubrication), and cause event C 12 (improper maintenance practices). This is because we use the wrong seal material, for example, elastomer, which is used in high temperature conditions, causes failure rapidly at the seal faces or interfaces before its operating life. Similarly, insufficient seal lubrication leads to increase in friction, and hence, the seal temperature increases. Similarly, improper maintenance practices also lead to increase in seal face temperature, for example, improper maintenance of springs leads to an increase in the pressure on seal faces, due to which rubbing increases, which further increases the seal face temperature. Therefore, the factors of improper seal material, insufficient seal lubrication, and improper maintenance practices, influence cause event C 10 (excessive seal face temperature).
Cause event C 6 (improper seal design) and cause event C 7 (problem of misalignment) are related to cause event C 10 (excessive seal face temperature) with a two-stage relationship. This is due to improper seal design, which leads to inadequate cooling at the seal faces, resulting in excessive seal face temperature. However, cause event C 10 (excessive seal face temperature) has a more immediate influence on cause event C 3 (improper seal material selection), cause event C 9 (insufficient seal lubrication), cause event C 12 (improper maintenance practices), than cause event C 6 (improper seal design) and cause event C 7 (problem of misalignment). This shows that no matter how these cause events lead to an increase in temperature beyond the permissible temperature limit, it leads to seal failure.
In a similar manner, the other non-zero elements in the reachability matrix can be interpreted in terms of various stage relationships among the cause events or between a cause event and seal failures. It may be noted that out of one-, two-, three-, and four-stage relationships obtained in the reachability matrix, that is, matrix (2), priority is given to the direct or single-stage relations, while suggesting appropriate action for their minimization. Thus mechanical seal failure reachability matrix (MSFRM), that is, matrix (2), shows a connectional relationship between two cause events and a cause event and the seal failure, which is then utilized for the importance of evaluation of a cause event.
| 6. Importance of The Cause Events|| |
The objective is to minimize the mechanical seal failure. This is possible if the stage relationships of the cause events of the mechanical seal failure reachability matrix (MSFRM) are evaluated. For example, let us consider column 6, that is, improper seal design in the reachability matrix, that is, matrix (2). It is observed that the cause events that are connected to cause event C 6 (improper seal design) by a single stage relationship are cause event C 3 (improper seal material selection), cause event C 4 (problem of maintaining 'PV' value), cause event C 5 (excessive fluid pressure), cause event C 9 (insufficient seal lubrication), and cause event C 10 (excessive seal face temperature). In fact, no matter how careful and skilled the design engineers are, their performance will be affected by the existence of any of the cause events.
Therefore, it is understandable that problems related to cause events, that is, improper seal material, problem of maintaining 'PV' value, excessive fluid pressure, insufficient seal lubrication, and excessive seal face temperature, are very important and crucial. The cause event C 12 (improper maintenance practices) is related to cause event C 9 (insufficient seal face lubrication) by the two-stage relationship. This means that improper maintenance practices lead to insufficient seal face lubrication, and hence, the importance of proper maintenance comes out for seal design. Furthermore, cause event C 7 (problem of misalignment) is related to cause event C 6 (improper seal design) by a three-stage relationship and can be interpreted in a similar manner. In this manner, the mechanical seal failure reachability matrix (MSFRM), that is, matrix (2) is useful for the important evaluation of cause events. In addition to this, all single-stage (i.e., direct) relationships between a cause event and the failure mode are obtained from [Table 1] or MSFCM, [i.e., matrix (1)] or MSFRM, [i.e., matrix (2)] and will lead to identification of the minimum cause event set, (i.e., M P ), corresponding to the failure mode.
For mechanical seal failure, this set is obtained as:
M P =[C 2 , C 3 , C 4 , C 5 , C 6 , C 9 , C 10 , C 11 ]
Even the complicated digraph can be conveniently analyzed by using digraph modeling and the matrix approach.
| 7. Conclusions|| |
The proposed digraph modeling as well as the matrix approach to the mechanical seals allows one to carry out their failure analysis. The significant feature of the present method is that unlike many other conventional methods, the approach is made not on a quantitative basis, but on a logical basis, using graph theory techniques. The procedure not only presents a better understanding of the relations of the cause events for the failures, but also helps in evaluating the stage relationships and their importance. Thus, it provides directions to designers and practicing engineers to take suitable corrective actions for the minimization of mechanical seal failures, thus improving the reliability of the machine and the machine elements employing them.
| References|| |
|1.||R. H. Warring, Seals and Sealing Handbook, 1 st Edition, Trade and Technical Press Ltd.; 1981. |
|2.||G. Lendaris, "Structural modeling - A tutorial guide", IEEE Trans Syst, Man Cybernatics, Vol. 12, pp. 807-840, 1980. |
|3.||F. Harary, "Graph Theory", Narosa Publishing House: New Delhi, India; 1988. |
|4.||F. Harary, R. Z. Norman, and O. Cartwright, "Structural Models - An Introduction to the Theory of Digraphs", John Wiley and Sons Inc.: NY, USA; 1965. |
|5.||R. Sehgal, O. P. Gandhi, and S. Angra, "Failure cause analysis of welded joints", IJEMS, Vol. 14, pp. 24-30, 2007. |
|6.||R. Sehgal, O. P. Gandhi, and S. Angra, "Failure mode analysis of rolling element", Proc Mechanical Engineering Congress and Exposition, ASME International: New York, Paper No. IMECE2001/DE-25115, pp. 1-7, 2001. |
|7.||A. D. Domashnev and G. V. Antipin, "Effect of operating condition on mechanical seal reliability", Springerlink Publication, Vol. 11, pp. 30-31, 1972. |
|8.||B. S. Zakharov, "Leakage in mechanical seal, chemical and petroleum engineering", Springerlink Publication, Vol. 43, pp. 1-2, 2007. |
|9.||N. Deo, Graph Theory, Prentice-Hall, Englewood Cliffs: NJ, USA; 1974. |
|10.||E. J. Henley and R. A. Williams, "Graph Theory in Modern Engineering", Academic Press: New York; 1973. |
|11.||I. J. Sacks, "Diagraph matrix analysis", IEEE Trans Reliability, Vol. 5, pp. 437-446, 1985. |
|12.||H. P. Alesso, "The relationship of diagraph Matrix Analysis to peri net theory and fault trees", Reliability Engg, Springerlink Publication, Vol. 10, pp. 93-103, 1985. |
|13.||R. G. Busacker and T. L. Saaty, "Finite graphs and networks - An introduction with applications", Mc-Graw Hill: New York; 1965. |
|14.||S. A. Warshall, "Theorems on Boolean matrix", JACM, Vol. 9, pp. 11, 1962. |
| Authors|| |
Mr. Jasbir Singh is working as an Assistant Professor in Department of Mechanical Engineering, N.I.T Kurukshetra. He has done B.Tech in Mechanical Engineering from Kurukshetra University in 2008 and M.Tech in Machine Design from N.I.T Kurukshetra in 2010. He has a teaching experience of more than 2 years.
Dr. Surjit Angra is presently professor in Mechanical Engineering Department at NIT Kurukshetra. He graduated in 1981; post graduated in 1986 and got Ph D. in mechanical Engineering in 1995. He has more than 25 years of experience in teaching, industry and research. His area of interest includes Tribology and Machine Design.
Dr. Vinod Kumar Mittal is working as an Associate professor in Department of Mechanical Engineering, N.I.T Kurukshetra. He has done B.Sc in Engineering (1989) and M.Tech. (1992) in Mechanical Engineering from R.E.C Kurukshetra and got Ph. D (2008) from Kurukshetra University. He has more than 19 years of teaching experience. He is a life member of several technical societies like MIE, TSI, ISTE and ISTAM. His area of research is Tribology and Machine Design.