Mechanical face seals are a good choice when minimum leakage of the sealed fluid is the most important criterion. The primary sealing interface is between rotating and stationary members that form a plane perpendicular to the shaft. The sealing area is a narrow ring where the two faces contact. One of the sealing faces is usually metal or ceramic and the other is usually graphite or plastic.
Standard face seals have been used for pressures up to 3,000 psi, rotating speeds up to 50,000 rpm, and temperatures from -425 to 1,200F. Special face seals have been developed for pressures up to 10,000 psi. For extremely high pressures, two or more face seals can be lined up in tandem, splitting the pressure differential equally.
Compared with other types of dynamic seals, face seals have longer life and can reduce warranty and liability cost, downtime, and leakage. In addition, face seals are easy to sterilize and eliminate system contamination by packing fragments.
Unbalanced face seals act like pistons. Fluid pressure from one direction loads the primary seal ring and mating ring against each other. Pressure from the opposite direction unloads -- and may even separate -- the rings. The result is either unnecessarily high friction, wear, heat generation, and power waste or else high leakage when the faces separate.
Unfortunately, there are also disadvantages to balanced seals. They typically cost 10 to 50% more than unbalanced seals for the same application because of closer tolerances and more complex seal shapes. They are subject to catastrophic failure if operating conditions do not closely match design conditions. And they usually require more space than unbalanced seals. Therefore, unbalanced seals are used whenever their frictional and pressure-reversal characteristics are acceptable.
Friction instabilities, such as stick-slip and oscillations in the sliding systems, cause detrimental phenomena that can generate positioning errors, poor surface roughness, noise and accelerated wear. In the automotive industry, many components could be affected by those undesired phenomena during deceleration regimes. The friction and wear behavior of mechanical face seals is ruled by lubrication conditions. Simulations based on tribo-dynamic models explain the occurring of friction instabilities during the operating conditions, describing different lubrication regimes: (i) full film or hydrodynamic lubrication regime, (ii) mixed lubrication regime and (iii) boundary lubrication regime. To avoid or limit instabilities it is fundamental selecting proper design parameters. Aim of the present paper is the set-up of a very fast and smart method to know how to reduce instabilities by tuning the correct dynamical parameters since the design phase. The proposed tool is based on ANNs that, even if it is not able to explain the frictional instability phenomena, as analytical models do, it allows to quickly investigate the ranges of parameters with respect to the operating range.
Hydraulic pumps are widely used in air and space vehicle fuel systems because of their advantages of high effectiveness and fast response. Failure of a hydraulic pump can produce serious consequences and the working environment of the pump is complex, creating enormous challenges for mechanical face seal design. Particularly in aeronautic and astronautic applications, the random vibration effect requires close attention. To achieve a high level sealing performance for mechanical face seals, an investigation of the influence of random vibration on mechanical seals is imperative to enhance the working reliability of hydraulic pumps.
The main components of a face seal are two rings, the stator, and the rotor, which are arranged perpendicular to the axis of the rotating shaft, as shown in Figure 1. The stator and the rotor keep fit under the force of fluid pressure and compensating mechanism. There are two degrees of freedom in the motion between the stator and rotor, which are the axial motion and the circumferential rotation. The contact between the stator and the rotor is the key to determining the performance of the mechanical face seal , and they can run in any of the following three lubrication regimes [7,8].
(a) Schematic of mechanical face seal; (b) sealing dam between the rotor and the stator; (c) schematic of asperity contact; (d) seal rings.
Various scholars have explored the dynamics of mechanical seals, including Greenwood and Williamson , who presented a contact model based on Hertzian theory. A plastic contact model was then proposed by Pullen and Williamson  a few years later. Dayan et al.  and Zou et al.  simulated the dynamics of mechanical face seals to prevent possible contact between surfaces, and Pustan et al.  presented theoretical models and experimental investigations of an internal mechanical seal with oscillating stator. The hydrostatic effect on the mechanical seals with oscillating stator was also examined. Minet et al.  analyzed and modeled the morphology of the mechanical seal face, and research by Leishear et al.  illustrated that relative vibrations between the stator and rotor affect the fluid film, damage the faces, and decrease the life of the seals. It was also determined that the smaller the axial damping between the two rings, the larger the impact force on the surface.
Random vibration has a significant influence on the axial direction of mechanical seals because it alters the lubricating regimes. Significant efforts have been made to study the random vibration responses of such systems [16,17,18]. Influence analysis under random vibration loading is more complicated than under conventional loading. Many methods have been put forward to facilitate such analysis, including stochastic averaging technique , the statistical linearization technique , and the equivalent nonlinear system method . Under vibration loading, mechanical face seals are in an unsteady state. Varney  analyzed the vibration of a noncontacting mechanical face seal caused by misalignment or imbalance. However, few authors have focused on exploring the influence of external random vibration loading on the mechanical face seals.
The focus of this study is the mechanical face seal used in air and space vehicles under random vibration loading. This study takes the surface roughness into account in the statistical sense from a macro perspective. As a small but unique contribution to this area of research, a spectrum analysis method based on frequency domain, and the root mean square (RMS) analysis method are presented to theoretically investigate the random vibration response characteristics. The research conclusions provide guidance for future robustness design under random vibration loading and to reduce wear in engineering applications.
Lebeck  proposed that the plausible value for mechanical contact pressure pc is the compressive yield stress SC. The contact interface portion of mechanical seal can be estimated by normal cumulative distribution function:
A schematic of the mass-spring-damping model is provided in Figure 2a. The mechanical face seal in the axial direction is abstracted as a mass-spring-damping system under random vibration loading. The housing is fixed to the pump and can be regarded as a moving base, while the O-ring produces the damping effect and the spring generates elastic force. A mass-spring-damping system with a discrete mass m, damping coefficient C, and spring stiffness K is placed on a moving base with acceleration y.
It is assumed that the mechanical seal is in a state of equilibrium in its initial state, that is, the closing force Fcl equals the opening force Fo. The purpose of this is to facilitate the study of response characteristics of mechanical seals under random vibration.
The acceleration power spectral density (PSD) of mechanical seals is shown in Figure 2b, as given by documents such as the MIL-STD-810G METHOD . In Figure 2b, ascending spectrum is from 20 Hz to f1, flat spectrum is from f1 to f2, and descending spectrum is from f2 to 2000 Hz. The values of f1 and f2, W1, W2, and W3 are obtained from specific tests. It is common to use the 3-sigma (3σ) values of the acceleration responses for verification of the mechanical structure design. The calculated xRMS(fn,ξ) value is equal to 1σ value. In this work, it is assumed that the mean value of random vibration response spectrum is zero and is Gauss distribution with a probability density of:
When the input acceleration PSDs are 4.0, 5.0, 6.0, and 8.0 (m/s2)2/Hz, respectively, the probability density of closing force Fcl under different input acceleration PSDs and the relationship with duty parameter G is shown in Figure 8. The green area represents full film lubrication regime, the purple area represents mixed lubrication regime, and the blue area represents boundary lubrication regime. When the input acceleration PSDs are small, the mechanical face seal mainly works under mixed lubrication regime, and with the increase of input PSDs, the probability of the mechanical face seal working under full film lubrication regime increases, with the probability value shown in Figure 9. Because of the balance radius rb and the initial preload of seal spring Fs, the mechanical seal will not work under boundary lubrication regime.
A simulation study was performed, and the results indicate that when the input acceleration PSDs are small, the mechanical face seal mainly works under mixed lubrication regime. With the increase of input acceleration PSDs, the probability of the mechanical face seal working under full film lubrication regime increases. The results also indicate that with the decrease of axial damping or the increase of axial stiffness, the probability of the mechanical face seal operating under full film lubrication regime increases.
Heat effects are the major cause of failure in mechanical face seals used in pumps, mixers, agitators and wherever a rotating shaft passes through the housing of a machine. Improper design, installation or maintenance of seals can lead to overheating and catastrophic failure of a system, endangering the safety of plant operators and posing an environmental hazard as a result of pollution associated with leakage from seals. 153554b96e