Printer Friendly

Reducing pressure loss of large diameter check valves.

Transcend Inc., a consulting firm that specializes in the use of computer simulation to optimize existing equipment and system designs, was approached by Mannesmann Demag AG, Monchengladbach, Germany to optimize the design of its DRV-B check valve. Mannesmann has been a world leader for over 70 years in the design and manufacturing of check valves for critical applications.

In one of the first applications of Computational Fluid Dynamics (CFD) technology to valve design, the pressure loss coefficient (K) of the DRV-B valve was lowered to 0.40-0.50 for valve sizes NPS48-NPS12, the lowest possible level for this type of valve. The flow efficiency is three times better than that of the earlier design. As a result, the optimized Mannesmann Demag DRV-B check valve provides a dramatic reduction in operating cost, particularly in transmission service where natural gas is transported over long distances.

The reduced pressure loss saves compressor fuel cost. For the optimized valve, the incremental compressor fuel cost is reduced to 1.5-times the capital cost of the valve calculated over a 20-year Life Cycle Cost (LCC) period. For a typical non-optimized vane, the incremental fuel cost is two to three times the valve's capital cost. Other improvements were a faster dynamic response and a reduction of the valve's capital cost. The dynamic performance was improved, through the installation of three-times stronger springs. The capital costs were reduced through a 25% reduction of the face-to-face length and several other design improvements.

Transcend used CFX CFD software from AEA Technology, Pittsburgh, PA in the optimization of the valve design. CFD was used to simulate the fluid flow through the valve and to analyze the effect of modified internal contours. The contour modifications were aimed at eliminating recirculation zones that are largely responsible for hydraulic losses.

Transcend worked with the Mannesmann design team to optimize the valve's performance. Optimization was based on the application of Transcend's curved wall diffuser design. The DRV-B non-slam check valve has major applications in natural gas, water and oil pipelines in sizes up to 60 inches in diameter. The prime operating function of a check valve is to close quickly at flow reversals to prevent damage to upstream piping and piping components. While performing this function, the valve should have minimum pressure loss during normal operation when it is fully opened. The standard method for measuring pressure loss involves measuring static pressure of two pipe diameters upstream and five pipe diameters downstream of the valve. The loss coefficient is then equal to the pressure loss divided by the dynamic fluid pressure.

Mannesmann's DRV-B check valve is a nozzle design with a dual annular diffuser and a ring-shaped disk. Valve closure is spring-assisted. The valve opens and closes at relatively low flow rates. Normal flow pushes the disc backwards and fully opens the valve. When the flow decelerates, the springs push the disk into the valve seat avoiding reverse flow and valve slam. In this type of design, flow accelerates in the seat area enabling valve opening while locally lowering the static pressure. The dual annular diffuser is subsequently used to gradually recover this pressure with minimum losses.

In general, pressure loss in this type of valve can be attributed to four principal sources.

Hydraulic energy losses arise from internal fluid friction and is irreversible. Turning losses occur when the flow changes direction around the valve disc before entering the diffuser. Mixing losses occur when the fluid streams of the two diffusers merge downstream. Losses associated with flow expansion occur in the diffuser as a result of flow separation and the formation of large flow recirculation zones.

Flow expansion losses account for the largest portion of the total pressure loss and are also the type of losses which the valve designer has the greatest power to reduce. If flow streamlines separate from the inner surface of the valve, recirculation zones will be formed that convert kinetic energy into potential pressure energy, resulting in pressure losses. The key to preventing flow separation from occurring is to increase the static pressure as gradually as possible along the length of the diffuser.

Until recently, valve designers had to rely on extensive trial-and-error testing of physical models. Improvements could only be monitored through the measured pressure loss across the test valve which gave little insight into the flow field through the valve. It was not possible to accurately determine the location and level of flow separation.

Transcend recognized the opportunity to improve the valve performance and to use CFD as a design tool. CFD involves the solution of the governing equations for fluid flow, heat transfer and chemistry at several thousand discrete points (cells) on a computational grid in the flow domain. The use of CFD enables engineers to obtain solutions for problems with complex geometries and boundary conditions. The CFD analysis yields values for fluid velocity and fluid temperature throughout the solution domain. As a result, valve designers can visualize exactly what is happening inside the valve and respond with engineering changes that directly affect problem areas. Another advantage of CFD is that alternative designs can be evaluated without the time and expense of constructing and testing multiple physical prototypes.

Transcend had several reasons for selecting AEA's CFX CFD software to model the valve. First, the software uses finite volumes (instead of finite elements) and provides very robust RNG K-epsilon and Reynolds Stress turbulence models that Transcend engineers have found to be significantly more accurate than other models they have evaluated in valve flow problems. Good turbulence models are key to accurately predicting the location of flow separation and the pressure loss across the valve. Second, CFX provides a powerful higher order convection scheme called CCCT. This convection scheme substantially reduces the number of cells required to solve a problem of this type. The computation time to arrive at a fully converged solution is approximately five times faster than with most other codes that only offer single order convection schemes. Third, the unique multi-block grid structure of CFX improves accuracy by avoiding the need for skewed element shapes. Finally, the AEA engineering staff offered timely technical support.

A CFD model of Mannesmann's original DRV-B design was created to serve as a validation case for the modeling setup. Taking advantage of its axi-symmetry, the valve was modeled in two dimensions in fully opened position at a flow rate of 2.5 m/s. The 60,000-cell model took about 10 hours to converge on an HP workstation with 128 megabytes of RAM.

Flow separation could be seen in the velocity vector plots as regions of low and negative flow rates relative to the inlet velocity. The analysis showed that the primary reasons for flow separation were poor flow conditions at the diffuser inlet and rapidly expanding cross sections in a straight walled diffuser. The original diffuser was not very effective despite the relatively small net diffusion angles. The analysis also showed secondary factors contributing to the pressure loss across the valve including high peak velocity at the seat angle, and the unequal and irregular velocity distribution and deceleration in the top and bottom diffusers. Transcend engineers concluded that redesign of the valve inlet and diffuser geometry had the potential to deliver a major reduction in the loss coefficient of the valve. Control of these two factors resulted in a very efficient high performance curved wall diffuser.

The flow profile of the valve was optimized through an iterative process in which the effect of every modification was evaluated using the CFD results to determine the next modification step in the process. Transcend applied its design approach for curved wall diffusers to avoid flow separation. The result is a valve with a pressure loss and dynamic performance significantly superior to competitive designs.

The calculated pressure loss from the CFD data showed excellent correlation with the prototype measurements. CFD analysis predicted a pressure loss coefficient of 0.52 for the optimized design. Physical testing of a prototype showed that this result was conservative and that the true value was about 0.50 (NPS12 valve). The optimized valve, as might be expected, has been a major success in the market place since its introduction.
COPYRIGHT 1997 Oildom Publishing Company of Texas, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1997 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Publication:Pipeline & Gas Journal
Date:Sep 1, 1997
Previous Article:Expert system aids reliability.
Next Article:Solution found to chronic trim damage.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters