Engineers at A.W Chesterton Co., Stoneham, MA, applied computational fluid dynamics (CFD) to a new model of their heavy-duty cartridge dual seal Chesterton's dual seal consists of two pairs of seal rings. Mechanical seals are widely used to prevent leakage from fluid-handling equipment such as centrifugal pumps and mixers. Outer pair rotates with the shaft; inner pairs are fixed and contain a channel for the barrier fluid. The seal confines process fluids to the areas on the left. Taper is visible along the lower edge of the inner seal rings and on the corresponding surface of the shaft covering. CFD images described in the article depict the axial circulation of the seal's barrier fluid for a typical untapered seal design and the improved circulation resulting from the tapered surface design. The change in flow patterns results in an improvement in heat removal, from 0.7 to 1.1 kW. In order to validate the accuracy of the CFD results, physical experiments were conducted in Chesterton's seal test laboratory, using a variety of flow rates, rotation speeds, and fluids. The seals that do incorporate the innovative design have performed well in the field, operating at cooler temperatures that should result in seal life at least 30 percent longer,


Mechanical Seals Are widely used to prevent leakage from fluid-handling equipment such as centrifugal pumps and mixers. In applications where zero leakage is important, such as in handling noxious or volatile fluids in chemicals, plastics, or pulp and paper manufacturing, dual seals add an extra measure of security.

Dual seals typically involve two stationary and two rotating rings and include a barrier fluid, which circulates within the seal and is maintained at a pressure higher than that of the process fluid. The barrier fluid also acts as a coolant to remove the heat generated when the stationary and rotating rings of the seal rub against each other. Frequently, water is the barrier fluid, but other substances, such as ethylene glycol or oil, can also be used.

One problem that has arisen with dual seals is that the cooling action is not always as efficient as it might be. To supplement lab testing, engineers at A.W Chesterton Co. in Stoneham, Mass., applied computational fluid dynamics to a new model of their heavy-duty cartridge dual seal. They wanted to study the circulation of the barrier fluid inside the seal in more detail than lab tests permitted.

Chesterton is a manufacturer of mechanical seals, mechanical packing and gasketing, and process pumps. Its products also include hydraulic/pneumatic sealing devices, maintenance chemicals, and ARC component materials.

Engineers at Chesterton expected that improving the circulation of the barrier fluid would increase cooling efficiency and extend the performance of the seal. Using computational fluid dynamics software to simulate the flow of fluid within the seal, they determined that the barrier/coolant fluid was not circulating well to areas of the seal where heat is generated. After a number of design changes were evaluated using computer models, engineers were able to improve axial movement of barrier fluid, and consequently heat removal, by almost 50 percent. Chesterton has now put the design change to work in a number of its products.


Rings, Fluid, and Friction

Typically, a mechanical seal consists of two sealing rings, one made of a soft material such as carbon graphite and the other of a harder material such as silicon carbide.

One ring rotates with the pump shaft while the process fluid is pumped. The other ring remains stationary. The interface between the two rings establishes the seal, preventing the process fluid from leaking.

When dual seals are used, the barrier fluid can be pumped from a separate tank thro ugh an inlet port into the seal. The fluid then circulates through the seal, to serve as a barrier and as a heat absorber. The fluid leaves the seal via an outlet port and flows back into the tank. The tank is cooled by either natural or forced convection.

The tank is cooled by either natural or forced convection.

Heat is often detrimental to the life span of a mechanical seal, yet friction generates quite a lot of heat, because the rotating and stationary seal rings touch each other. If the seal is not cooled adequately, heat can distort the rubbing surfaces, causing the unit loads to concentrate in a few areas rather than distribute uniformly over the entire interface. Concentrations of pressure can cause excess wear and shorten the life of the seal.

In addition, the elastomeric O-rings used to seal the rings to other component parts must allow the rings to move slightly if the seal is to perform properly. The O-rings perform best at cooler temperatures and tend to harden if they become too hot. If they become hard, they can prevent the necessary motion for the seal rings and ultimately lead to leakage.


Problems with Design

The first step was to understand exactly what was happening with the coolant within the existing seal. This sort of insight was impossible to obtain from routine laboratory testing, which can provide the temperatures of the fluid only at the inlet and outlet ports. To learn temperatures inside the seal and internal flow patterns, engineers turned to CFD simulation. The software used for the e FD analysis was Fluent from Fluent Inc. in Lebanon, N.H.

Engineers simulated a seal for a 48-t11m-diameter centrifugal pump shaft rotating at 3,600 rpm. The initial operating conditions were assumed to be 687 kPa and 66°C for the pump process fluid and 1,031 kPa and an inlet temperature of 38°C for the barrier fluid. The numbers fell within the middle range of operating conditions for this design; the simulated operating conditions were then varied to some extent. The variables studied included the radial clearance between stationary and rotating boundaries, taper angle of flow control sl1lfaces, shaft rotational speed, barrier fluid through flow, and key thermophysical properties of the fluids. The Fluent simulations involved three-dimensional models with approximately 160,000 cells, and were carried out on a two-processor machine. Convergence of the analyses typically required around 2,000 iterations to reach the desired level of accuracy. Postprocessing of the simulation results was performed using IBM's Visualization Data Explorer software. All the erD and data visualization analyses were run on Silicon Graphics workstations.

The results of the simulations clearly indicated how the seal cooling could be substantially improved. Barrier fluid enters the seal at the inlet port via a flow channel, which is axially centered between the heat sources. Cooling efficiency, therefore, is dependent 011 how well the fluid circulates axially to the heat-producing areas where the seal rings meet. A graphic presentation of the CFD results, consisting of color-coded vectors indicating fluid tmperature, as well as velocity magnitude and direction of flow revealed limited circulation to these critical regions of the seal.

The next step was to find ways to improve axial circulation of the barrier fluid. CFD simulations were used to evaluate the effectiveness of each modification considered. The change found to have the largest effect on axial circulation was tapering the bounding surfaces of the stationary seal rings and the shaft sleeve. Axially tapered surfaces propel cool fluid from the flow channel toward the heat-producing regions of the seal. Compared with the original design, the tapered surfaces were shown to be far more effective at promoting axial flow. As expected, increasing axial flow resulted in better heat removal. The modified design showed an in crease in heat removal about 50 percent better than the original configuration—1.1 kW heat removal for the new design versus 0.7 kW heat removal for the old.

Engineers also used CFD to determine the physical mechanism responsible for the improved performance of the tapered surface design. Simulation results showed that the fluid near the shaft sleeve experiences a strong centrifugal force directed radially outward from the center of rotation.

In the case of the tapered-surface design, this radial load has an axially directed component that drives fluid away from the flow channel toward the ends of the domain where fluid sealing and heat generation occur. A graphical presentation of the turbulent kinetic energy of the flow showed regions of relatively high turbulence near the sealing interface and near the inner, rotating wall, closer to the flow channel. This situation is advantageous since the higher turbulence and increased mixing help promote heat transfer where it is needed most.

To better understand the nature of the axial exchange of fluid and associated thermal energy, trajectories of fluid particles released near the flow inlet were computed and displayed as stream ribbons colored by fluid temperature. The flow trajectories were then animated, with twisting ribbons indicating the local level of turbulence, or vorticity. The animation provided a detailed three-dimensional perspective of the helical flow patterns characteristic of the tapered-surface design.

One of the techniques used allows the observer to travel the route of a fluid particle. The presentation displays the local peed and temperature of the particle, as well as the elapsed time from its release. At about 55 milliseconds into the animation, the temperature-mapped trajectories, induced by low pressure, can be seen bypassing the outlet en route to the interface regions of the seal. About one-twentieth of a second later, one of the paths of the accelerated particles traces alongside the rotating radial end wall of the domain. At this point, the particle has absorbed enough heat to result in a temperature rise of 12°C. After absorbing heat from the sealing interface, it reaches 62°C, the maximum temperature attained along its path. During the return trip to the flow channel, the speed of the particle drops by about 25 percent as it traverses its way near the inner boundary of the stationary seal ring.



Checking out the Results

To validate the accuracy of the CFD results, physical experiments were conducted in Chesterton's seal test laboratory, using a variety of flow rates, rotation speeds, and fluids. The temperature of the fluid was measured as it entered and left the seal, and these data were then compared with the computer analyses . A representative case showed the predicted fluid temperature in the flow channel region for the tapered surface design rising by approximately 11°C from the inlet to the outlet. The corresponding lab data indicated a temperature rise of 10°C, within 1°C, or 10 percent, of the CFD simulation.

This work led to design changes in some of Chesterton's products. It was not possible to implement the tapered surface design on all of the firm's mechanical seals because a certain amount of radial space is needed to provide the taper. Some pumps don't have enough space to accommodate a sufficiently large seal.


The seals that do incorporate the new design have performed well in the field , operating at cooler temperatures that should result in seal life at least 30 percent longer.

They have also done well in the market, quickly establishing themselves as one of the firm's three or four top product lines, in what is generally a highly conservative industry. And the company's patent application has 6nally gone through. A patent was granted on the new seal design last August 17.