How NOT To Use a Rubber Expansion Joint
How NOT to Use a Rubber Expansion Joint
What’s wrong with this picture?
By: Marty Rogin
Rubber expansion joints are likely the least understood and most abused component in a piping system. They are flexible, stretchy, bendy and easily forced into lots of places despite what the installation instructions say. Most of the time, rubber joints are merely an afterthought in multimillion-dollar piping systems. Until things go awry.
The rubber joint is unmatched for vibration isolation. Properly installed, a rubber joint will greatly reduce equipment nozzle loads. Its resilience allows it to be installed in many different systems under a huge range of temperatures, pressures and media. “What could possibly go wrong?”, you may wonder.
Blame Mr. Murphy if you want, blame human nature, the Fates or the alignment of the planets. The reality of most failures is more straightforward. Most of the time, it is installation. More specifically, not following the manufacturer’s instructions. What follows is a rogue’s gallery of photos illustrating the ugly aftermath of ignored installation instructions and unforeseen operating conditions. Learn these lessons well so your piping system does not become a subject of another article.
Sometimes flexibility is a disadvantage. Why? Because it’s easy to compress a joint into a space that’s too small, which is exactly the problem here. The bead was damaged as the joint was forced into a gap between flanges, resulting in a seal failure. Spherical expansion joints rely on this bead to form a seal between flanges. If the bead is damaged, the building engineer will curse your name for eternity. Don’t violate the face-to-face dimensions of an expansion joint.
- Joint installed between two misaligned flanges.
- Joint begins leaking at the flange-to-flange seal in a week (or month, or several months…).
- Bolts tightened, leak stops. In the meantime, the rubber bead takes a compression set becoming less resilient.
- Repeat steps 2 and 3 several times until…
- Bead is compressed to about 1/16th inch, rips apart from the body, pump room is now a water park.
Don’t turn your pump room into a water park – or even worse, a sewage tank. Align those flanges before installing expansion joints.
Did you know your water pumps can generate steam? This operator didn’t. In this unfortunate scenario the operator closed the pump isolation valves with the pump operating, dead-heading the pump. This situation is ok for a short duration, but eventually all that mechanical energy added to the water has to go somewhere. It went into heat. The water contained in the pump and pipe up to the isolation valves had so much energy added that it flashed to steam. The expansion joint was the first component to fail, which was fortunate for the pump. The temperatures and pressures exceeded the rubber performance limits and the joint failed, nobly sacrificing itself for the greater good of the pump and piping.
The previous example showed both temperatures and pressures out of limits. If only the temperature exceeds the rubber rating, the joint will still fail but it won’t be quite as spectacular. This photo illustrates a hard-boiled expansion joint. Rubber turns hard and brittle when exposed to temperatures exceeding the published limits. It won’t be immediate, but over time the rubber will essentially transform into something resembling plastic. And it will no longer move. This is bad.
Vacuum sucks. Literally. A vacuum is any point in the pipe where the pressure drops below atmospheric pressure (14.7 psia/29.92” Hg). There are some expansion joints that can deal with a partial vacuum, and some that can’t. This unfortunate joint can’t. This poor spherical joint was subjected to a vacuum and over-extended itself. In building systems, expansion joints can be exposed to unintentional vacuum conditions when a riser is drained and not vented, or if a pump inlet pressure is below atmospheric pressure. Always check those operating conditions, then follow the manufacturer’s directions for installing your joints for the conditions!
Material compatibility issue? When your expansion joint turns to goo, that’s a very good first guess. This is the most unfortunate aftermath of the media reacting with the rubber. This situation could have been avoided by selecting a different elastomer.
Absent from these photos is the expansion joint’s trusty sidekick, the control rods. Control rods are not necessarily required for every installation, but they are always a good idea. Consider control rods as cheap insurance, preventing expansion joints from over-extending or over-compressing. A control rod assembly will avoid problems related to excessive movement.
One more item the author has encountered at EVERY site where a rubber joint failed is loose bolts. This is no exaggeration. EVERY failure site had expansion joints with bolts that could be removed without tools – even on joints that were not involved in the failure. Although manufacturers have different bolt tightening requirements, the one common thread (bad pun intended) is that bolts must be re-tightened after the initial installation.
What do all these failures have in common? It all boils down to human error. Each incident could have been avoided by following the manufacturer’s instructions and effectively communicating the application requirements prior to ordering the expansion joint. Convincing people to follow instructions can be a futile effort. Communicating application requirements, by contrast, is easy. Appendix L, p.49, of the Piping Expansion Joint Handbook has one single information sheet that can eliminate many of the problems shown in this article.
The Fluid Sealing Association’s Piping Expansion Joint Handbook and KnowledgeBase contain a wealth of information, so consult these resources early and often during the project.
How gravity, volume and thermodynamics play a role in smart riser design.
By Marty Rogin, PE; Engineering Manager, Metraflex
The modern skyscraper has been around for over a century. Like other elements of our built environment, the skyscraper can only exist because of other innovations in building technology, namely steel frame construction and safe elevators. Even though we have figured out how to build strong, tall structures and safely move the people inside, there are still the challenges of heating and cooling the building, moving fresh water in and dirty water out, providing fire protection and electricity. Defying gravity adds another twist to the challenges of providing services within tall buildings. This article will introduce some basics of pipe riser design and performance, explain some considerations in using different expansion joints in pipe risers, and briefly describe some of the codes and standards regarding guiding and supporting risers.
Thermal Expansion Basics
While the pipe is nothing special, gravity will make things way more interesting. Consider the riser pipe (Figure 1). The pipe runs the entire height of the building, 50 stories. If the slab-to-slab height is 10 feet, our pipe is 500 feet tall. A typical support for this pipe may be a riser clamp, maybe on every other floor. With no temperature change, the riser weight is distributed evenly between all the riser clamps.
Let’s heat the water in the pipe (Figure 2). Now the pipe will expand against the supporting riser clamps. But the riser clamps are only restricted to move in one direction – down. There are no restrictions to upward movement. The clamps will just move up with the pipe. Any clamp above the bottom floor will now be floating above the slab. All the weight of the pipe, insulation and media is on the bottom clamp. Most pipe clamps are not designed to support the full weight of a tall riser.
There are solutions. A pipe anchor at the bottom of the riser, designed to support the full riser weight, will solve this problem. But let’s look at how much the pipe moves. Let’s say our pipe is made of steel and the liquid medium is hot water at 180°F. Like gravity, thermal expansion (thermal strain) of steel will not disappear in a riser. If we assume an ambient temperature of 50°F, the pipe will want to expand according to the equation:
ΔL = Length change (inches)
∝ = Coefficient of thermal expansion (for steel, 6.33×10-6 inch/inch/°F)
Lo = Starting length (6000 inches)
ΔT = Temperature change (180°-50° = 130°F)
ΔL = 4.9 inches
The very top of the riser will move 4.9 inches up. Is this a problem? It could be. Can the takeoffs at the upper levels move about 5 inches without breaking? Maybe, if there is enough runout length to the equipment connections. Will the field conditions allow the pipe to move this much before colliding with structure or equipment? Maybe, but then, who can answer these questions prior to construction? Usually they can’t be answered until the structure is up and the pipefitters install the pipes at the ceiling with all the unplanned bends and modified runout lengths.
One solution may be to move the anchor to the center of the riser (Figure 3). The anchor is a hard connection from the pipe to the structure and a point of zero movement. The riser is now divided into two sections, each 250 feet. Now the maximum pipe movement will be half of the entire riser, or 2.45 inches. The previous questions may be asked regarding 2.45 inches of movement. If they can be answered during the design stage of a project, great! On to the next project!
But wait. What about those riser clamps? Above the anchor they will ride on the pipe, rising above the floors. But below the anchor, the riser clamps will try to restrain the pipe from moving downward. The likely outcome will be that the clamps will slip along the pipe as it moves. If the riser clamps are welded to the pipe, something will break – either the clamp or the pipe. Hopefully the clamp, but then the anchor will be carrying the load of the entire riser.
Riser Spring Supports
And what about spring supports? These are specially-designed systems of anchors, guides and supports for risers that are designed to move with the pipe. The spring supports stay in contact with the floor slab as the pipe moves. As the pipe moves, the springs stretch or compress to exert more force on the floor slab, which takes load off the main anchor in the center of the riser. These systems are effective for taking the load off the main anchor; however, this type of system has limitations. These are:
- The pipe still moves! Nothing will prevent this. If we use our 500 foot riser as an example, the anchor would be in the center, and the ends would move the same 2.45 inches.
- Only one anchor is permitted in each riser. A second anchor will restrict the pipe movement, resulting in tremendous forces in the anchors and floor slabs while adding potentially huge stresses in the pipe.
- It is unclear if this type of system can be adapted to copper risers. The available manufacturers’ literature does not specifically mention copper as an acceptable pipe material for these support systems.
A riser system using pipe riser clamps or spring supports will have limited control on the pipe movement. Expansion joints allow for better control of the pipe movement. Before looking at expansion joints, let’s consider what happens to the internal pressure of a riser.
Pressure and the Height of a Water Column
The internal pressure along a horizontal pipe axis generally varies a small amount. Once that pipe is tipped up to vertical, a fluid-filled riser builds pressure as the pipe gets taller. The pressure at the bottom can be significantly higher than at the top. This is due to the weight of the water.
Consider a tank with 1 foot of water (Figure 4). No matter how full, the tank will experience more force on its walls towards the bottom. The most force will be at the bottom of the tank. Each added inch of water in the tank increases the weight that the bottom of the tank must hold. When the height of the water reaches 27.7 inches, there is 1 pound on each square inch of the bottom of the tank (Figure 5).
Now, let’s change the shape of the tank to something narrower (Figure 6). As we move the tank walls closer, we need less water to fill the tank to 27.7”, but the tank bottom has a smaller area. The force on each square inch is still 1 pound.
It doesn’t matter what shape we make the tank, or even if it’s a pipe; if the water column is 27.7” tall, the pressure at the bottom is 1 psi.
If we stack these 27.7” water columns, the pressure at the bottom builds in 1 psi increments (Figure 7).
The pressure at the bottom of the stack increases by 1 psi for each 27.7” section. Conversely, the pressure increases by 0.43 psi for each 12” section of water. Using this logic, the pressure at the bottom of our 500 foot riser that is only due to the height of the column of water will be:
This is referred to as hydrostatic pressure, and it is why hydronic equipment is seldom located in the basement of a tall building. This is also why very tall buildings have risers that are subdivided between intermediate mechanical equipment rooms. For steam, gas and air, column height is not an issue due to the much lower density of these substances.
Riser Structural Stability Considerations
Column buckling is a familiar failure mode. If a long, slender bar is subjected to axial forces at each end, it will bow out (Figure 8). This is a function of the material strength, cross section dimensions and length of the bar. A pipe behaves like this too. Axial forces applied to the pipe ends will also make it bow out. This can be especially pronounced on small-diameter copper pipe.
Although most of this bowing is elastic, meaning the pipe goes back to its original shape after the loads are removed, this can be a problem if the pipe bows beyond the elastic limit of the material. Column buckling can also be a problem with bellows expansion joints. If the two ends of a bellows are not within the offset movement limits, the expansion joint will be permanently damaged.
The pipe must remain aligned as it travels through the building. This is the purpose of pipe guides, which restrict the pipe to move only in the axial direction and essentially make the pipe more rigid. Guides divide the pipe into shorter, stiffer sections.
The spacing of pipe guides is dictated by the classical column buckling equations, called the Euler buckling equations. If we assume the pipe is pinned on either end, the equation looks like this:
This is the theoretical load limit for a column with the ends free to rotate and loads applied along the column axis. Notice that the weight of the pipe and water are not considered here. Euler buckling is an important consideration when bellows expansion joints are chosen for a piping system, especially risers as the forces are now acting along the pipe’s longitudinal axis.
If the pipe is fixed at one end (Figure 9) the critical load is:
What happens when the pipe is turned up on its end? Gravity. The weight of the pipe and media inside the pipe will now figure into the calculations. A riser pipe can theoretically collapse under its own weight (Figure 10). The critical load on a vertical pipe with the end fixed is:
Using the 4” pipe as an example and solving for the length with (ql)cr equal to 1.34 lb/in, the maximum length a vertical 4” sch. 40 can be is about 90 feet before becoming unstable. For comparison, a 4” type K copper riser will become unstable at about 64 feet. This is also the equation that determines the maximum height of a tree (neglecting the branches and assuming a prismatic trunk).
Next, consider a riser having an external force like a bellows pressure thrust and spring force. A riser pipe under an external load subject to the weight of the pipe wall and media inside will have a critical load of:
This equation assumes the end of the pipe is fixed and can’t rotate, the pipe has a constant cross section (same size all the way up) and that the weight is equally distributed. The critical load is reduced by 30% of the column weight. Note that the critical load can be negative, meaning that the top end support must be in tension to prevent buckling.
The previous examples along with the hydrostatic pressure explanation are important for guide spacing in risers with different types of expansion joints. Let’s first consider a bellows expansion joint in a tall riser. How would we determine the pipe guide spacing for this type of installation?
What are Pipe Guides?
Pipe guides are devices that allow the pipe to move axially, while restricting the pipe from moving perpendicular to the pipe axis. By restricting the pipe to only axial motion, the pipe is more rigid and will not bow out or collapse. As the guides are placed closer along the pipe, the amount of axial loading can increase before the pipe becomes unstable.
Common guides used for HVAC and plumbing systems are either finned or sliding. Finned guides, or “spider” guides, have fins fixed to the pipe and travel through a ring secured to the building structure. These guides are typically found on small-diameter pipe and are used for areas where the lateral loads are expected to be relatively small compared to the pipe anchor loads. In horizontal apaplications, these guides are not intended to take the place of hangers, so a clevis or roller support would be required in the vicinity of the guide to hold the weight of the pipe.
A more robust guide that can also function as a support is the sliding guide. This device has a sliding bar welded to the pipe, with a base secured to structure. The base has either Teflon, graphite or an elastomeric pad to reduce friction. This type of guide can handle greater lateral loads and is typically used on larger-diameter HVAC pipe or process piping. A version of the sliding guide adapted to risers incorporates an elastomer cushion between the slide and base to dampen noise and vibration of the pipe sliding against the slab penetration.
The most compact guide configuration consists of an elastomeric seal assembly within the slab penetration to guide the pipe. These take up no floor space in the riser chase and allow for the most efficient space usage.
Standards for Guide Placement with Bellows Expansion Joints
According to the Expansion Joint Manufacturer’s Association (EJMA) standards, guides are required with bellows expansion joints at a maximum distance of four pipe diameters from the joint, then a maximum 14 pipe diameters from the first guide for the next location. Subsequent guides are spaced at intervals dictated by the Euler buckling equation for a half pinned-column. When guides are placed according to EJMA guidelines, the pipe is subdivided into rigid sections that shouldn’t (theoretically) buckle under a known end load.
Guides with bellows expansion joints serve two purposes; to keep the pipes from buckling, and keep the bellows from squirming (Figure 12). The EJMA standards assume a horizontal pipe, and the buckling formula used divides the calculated length in half. For comparison, the 4” steel pipe with a bellows expansion joint under 158 psi requires an intermediate guide spacing of 30 feet. The pipe is assumed to be horizontal, so the weight of the pipe and media are not considered in the EJMA calculations.
Typical model codes require risers to be supported roughly at every floor. This is usually accomplished with riser clamps. As described previously, the riser clamps may move upward and lose contact with the floor slab, depending on the anchor placement. Now the support is not doing its job, and all the load is carried by the anchor. In this case, the codes were followed but the anchors may not be designed for the entire weight of the pipe, insulation and its contents – plus any forces generated by the expansion joints.
Bellows Expansion Joints in a Riser
Bellows expansion joints in risers are very common, mostly due to their compact shape (Figures 13 & 14). They take up very little space perpendicular to the pipe axis, so they fit nicely in crowded pipe chases; however, they do need to be guided. The bellows will produce large anchor loads. This may be a necessary trade-off, as space in pipe chases can be at a premium.
Pipes in the vertical are now subject to hydrostatic pressure variations. These variations are simple to calculate, and will vary from the system operating pressure at the top of the riser to the height divided by 2.31 added to the system pressure at the bottom of the riser. Using the 500’ riser as an example with a system pressure of 50 psi, the riser top will be at 50 psi and the bottom will be at 267 psi. This difference in pressure is critical in calculating the anchor loads for a bellows expansion joint.
A bellows expansion joint installed near the bottom of a tall riser must be rated for the pressure at this location. In the previous example, a 150 psi expansion joint would be fine for the top section of the riser, but a joint near the bottom would require a higher pressure rating.
And what about the anchor loads? Bellows expansion joints create reaction forces based on two characteristics of the bellows; the spring rate and the effective area. The spring rate is simply the amount of force required to compress or extend the bellows one inch. If a bellows has a 500 lb/inch spring rate, it will exert 500 pounds on each anchor for every inch of movement. If the bellows is compressed 1.5 inches, the spring force will be 750 lb. on each anchor.
Pressure thrust may not be quite as straightforward. An expansion joint is the most flexible part of the piping system. It has to be this way. A bellows under pressure wants to stretch back to its original shape, which is a tube. If left unrestrained, a bellows under pressure will extend past its rated movement. This is why control rods and anchors are generally required for a bellows expansion joint. The amount of force exerted by the bellows on either the anchors or the control rods is also simple to calculate. It is the pressure multiplied by the bellows effective area.
And what exactly is a bellows effective area? It is the inside area of the bellows, calculated at the average of the largest and smallest convolution diameters. This is also called the mean diameter. All bellows manufacturers provide the effective areas, so it is not necessary for the specifier to calculate this.
If we use our 500’ riser as an example, a bellows expansion joint at the very top of the riser with a system operating pressure of 50 psi and a 4” pipe (with a 4” expansion joint) will have a pressure thrust on each anchor of:
If we decide to split the riser and locate an expansion joint at the midpoint, the pressure used to calculate thrust force will be 50 psi added to the height of the water column above the expansion joint (about 250 feet):
Now let’s add in the spring force. The riser will move 2.45 inches between each set of anchors. If the spring rate of the bellows is 200 lb/in:
Friction from the pipe supports may be assumed to be very small for a riser, and will not be factored into these calculations. The total bellows force on the anchors will be:
What about the weight of the pipe, water and insulation? This must be added to the bellows anchor loads to get the complete picture. And the lower bellows forces act upwards on the middle riser anchor, while the upper bellows forces act down on the anchor. It is important to not only keep track of the magnitude, but also the direction of the forces acting on an anchor. Additionally, the anchor is carrying the weight of the pipe and water from above. The intermediate anchor loading is more complicated if the expansion joint is centered between the anchors.
We now have a situation similar to the critical loading for a riser under its own weight with an external force. If we look at our critical load equation (4) with pipe weight,
and solve for the length using Pcr = 6178 lb, the guides would require spacing at 23 foot intervals, or maybe every other story.
If a copper riser is installed, more guides would be required. The bellows forces would be roughly equal, as would the hydrostatic pressures. If the bottom half of the riser is once again considered, the only difference will be the material and cross-section properties of the copper pipe. For our 4” riser, the copper material and section properties are:
Now the required guide spacing is at 12.5 foot intervals, or maybe at every story.
Flexible Hose and Braid Loop Expansion Joints in Risers
The only way to really limit the amount of movement in a riser is an expansion joint. Movement can be limited to any acceptable amount by anchoring the riser at various levels and installing an expansion joint between each anchor pair.
Hose and braid expansion joints are another option for risers that provide many advantages over bellows expansion joints or spring support systems. Hose and braid expansion joints are typically constructed of two pieces of corrugated metal hose wrapped by metal braid. The joint may be fabricated in a ‘U’ or ‘V’ shape, which provides movement in all directions. Like the other expansion joint systems, hose and braid expansion joints are life-of-building products. They require no maintenance or inspections once installed.
Hose and braid expansion joints offer several advantages over bellows or spring supports:
- No pressure thrust component. This is due to the hose and braid configuration and the braid restraining the hose from expanding.
- Hose and braid expansion joints can be designed for operating pressures commonly found in HVAC and plumbing system pipe sizes.
- Hose and braid sections are very flexible. The only anchor forces generated by these expansion joints are due to the spring forces of the hose and braid, which are typically less than 100 pounds for many pipe sizes. The only other load on the anchor will be the weight of the full riser.
- Hose and braid expansion joints can handle offsets in the riser much better than bellows expansion joints.
Figures 15 and 16 show examples of hose and braid expansion joints commonly used in risers.
The only potential disadvantage of hose and braid is the space requirement. Bellows expansion joints fit nicely in crowded pipe chases, hose and braid sticks out. Even this situation can be accommodated by mounting the loops horizontally in a ceiling chase.
Hose and braid expansion joints subject the risers to small reaction forces, so the bottom half of a riser between anchors may be considered a free-standing length of pipe, similar Figure 10. This configuration would then follow a variation of equation (3) for the portion of riser below the expansion joint:
The term (q) is known, so the length for column stability (and guide spacing) can be determined by solving for the length l. Going back to the original example of a 500 foot, 4” sch. 40 pipe with a hose and braid connector in the center, the lower half would be subject to similar conditions as a riser of 250 feet with a fixed bottom. The required guide spacing would then be 10.6 feet. For type K copper, the required guide spacing would be only 4.1 feet.
For the portion of pipe above the loop, one guide at the expansion joint would be adequate. Gravity is working in a favorable direction in this case.
How often are guides placed every other story, let alone every single story in a high rise? Almost never. So why do we not have risers collapsing on every project? The answer may be simple; there are already guides at every floor, in the form of round slab penetrations. These allow axial movement and restrict lateral movement. Guide placement would be critical in an open chase, where pipes are routed through one single large floor penetration at each level.
Also, most risers have takeoffs or runouts at each floor. If these are hard-piped to equipment in the vicinity of the riser, this arrangement can provide additional lateral support to the riser.
Referring back to our original example of 4” steel pipe, EJMA guidelines don’t cover the case of a vertical pipe with zero loading (like hose and braid), and for the bellows loading in this example, recommend a 31 foot spacing (or about every three stories). This author has observed exactly zero riser installations that comply with EJMA guidelines for guide spacing, and has yet to see a collapsed pipe riser.
“Out of sight, out of mind” may also be part of the issue. The pipes may very well be bowing elastically, but nobody can see it. After all, how many architects will design windows on pipe chase walls? For that matter, how many tenants really care to observe their risers?
Although standards and codes address risers and guide spacing in pipes with bellows joints, it is important to know the limitations of the equipment and assumptions used to arrive at the recommended standards. Perhaps it’s appropriate to take a closer look at these standards and adapt them for tall risers.
Building utilities have to be distributed to all levels, or there would be no point to a skyscraper. Assuredly, as long vertical pipes are placed inside tall buildings gravity will always act down, and building system designers should be aware of the forces on these elements. The petroleum industry is well aware of the design considerations on tall, flexible risers through experience with offshore drilling rigs. As we build taller structures, the A/E/C community must also be aware of similar – but not identical issues for conditions above the surface.
Sparkes, C.P., Fundamentals of Marine Riser Mechanics, PennWell Corp., 2007
Timoshenkos, S. and Gere, J., Theory of Elastic Stability, McGraw-Hill, 1961
Flow performance tests were conducted at the Utah Water Research Laboratory (UWRL) on 3-inch, 4-inch, 6-inch, 8-inch and 10-inch Metraflex strainers. In addition, 4-inch, 6-inch, 8-inch and 10-inch Competitor strainers were also tested. The testing was developed to determine the flow versus pressure loss characteristics of each strainer and the flow coefficient of each strainer. Each strainer, with the exception of 3-inch Metraflex strainer and the 4-inch Competitor strainer, had strainer perforations measuring 0.125 inches in diameter. The 3-inch Metraflex strainer had 0.045 inch diameter perforations and the 4-inch Competitor strainer had 0.033 inch diameter perforations. The work was authorized under Metraflex Purchase Order No. 29582 and was done in accordance with the ANSI/ISA 75.02.01-2008 Control Valve Capacity Test Procedures standard with slight modifications in order to characterize each strainer’s performance over a wide flow range.
Each of the strainers was installed in a test line (standard steel diameters) with approximately 20 diameters of straight approach piping to provide uniform flow at the inlet of the strainer. There were approximately 10 diameters of straight pipe downstream from the strainer. Pressure taps were located two pipe diameters upstream from the strainer and six pipe diameters downstream from the strainer. Flow was supplied with a 100 horsepower pump. Figures 1 through 4 show the test setup for the 4-inch and 10-inch strainers respectively. Each of the other sizes had similar installations however they were installed in pipe sizes corresponding to each strainer size.
The flow rate was measured using calibrated flow meters which were verified against certified weight tanks. The differential pressure across the strainer was measured using Rosemount differential transmitters. The upstream pressure was measured using a Rosemount transmitter. The water temperature was measured using a calibrated RTD.
Each strainer was tested over a range of flows sufficient to generate 1.5 ft/s to 15 ft/s average velocities in the approach pipe. Ten points were taken over the velocity range of each size.
In the case of the 6-inch Metraflex strainer, certain tests were completed to simulate strainer blockage. This was accomplished by using duct tape to completely block the screen in areas suspected to plug first. The plugged portion of the screen was side opposite the strainers inlet.
To prevent screen plugging during testing, Metraflex provided a 10-inch 20 mesh strainer that was used upstream to capture any debris that entered the test system from water supplied from Logan River
The definition of the flow coefficient used in this report is:
Where Q is the discharge of test fluid in U.S. gallons per minute flowing through the strainer, ΔP is the pressure drop across the strainer in psi, and SG is the specific gravity of the test fluid. Cv is calculated using the gross pressure drop (ISA standard) between taps that are two diameters upstream and six diameters downstream.
The net flow coefficient was also computed by subtracting the friction expected from steel pipe between the gross differential pressure measured between the pressure taps. This calculation was completed using the Swamee-Jain equation to determine the friction factor associated with the specific data point taken in the laboratory. The Swamee-Jain equation is given by:
Where f is the friction factor, k is the pipe roughness, D is the inside diameter of the pipe and Re is the Reynolds number of the flow in the pipe. The headloss in feet was then converted to psi and subtracted from the gross pressure loss measurement.
The test procedure essentially followed ISA 75.02.01-2008 with slight modifications to account for the fact that a strainer is not a valve and it was desired to determine the strainer’s performance characteristics over a wide flow range.
- Install the strainer in straight piping of nominal size and standard wall thickness. Ensure that at least 20 diameters of straight pipe are upstream from the strainer and at least 8 diameters are installed downstream from the
- Flow test the strainer at several different flow rates and observe the relationship between flow and Cv.
- The following data shall be recorded:
- Upstream pressure (measurement not to exceed 2 percent of actual value).
- Pressure differential across the strainer (measurement not to exceed 2 percent of actual value).
- Volumetric flow rate (measurement not to exceed 2 percent of actual value).
- Fluid temperature (measurement error not to exceed 2 degrees Fahrenheit).
- Strainer description and identifying
- Calculate the gross and net flow coefficient Cv.
The pressure loss and flow coefficient Cv data are given in Tables 1 through 9 and the net pressure loss and net flow coefficient data are shown graphically on Figures 5 through 14.
Metraflex manufactures expertly engineered products designed to handle a wide range of movement in any application. (more…)
Shortly after Metraflex introduced the first flexible pipe loop (the Metraloop), it became the standard used by engineers and designers to protect piping systems in seismic events, exactly what you would expect from something invented by a person named Jim Richter. Metraflex has been leading the way with seismic solutions ever since.
All the Metraloop, V loop, and Dog leg products are all ideally suited to be used for seismic applications. The low spring rates, movement in all directions, and that they do not impart any thrust loads on the piping reduce loads on the seismic bracing make them ideal for seismic applications. Additionally, these products are capable of movement in all six planes.
All these products are available in standard configurations and customized versions, tailored to meet your projects requirements. Additionally, the Seismic Gator was developed to be used when there was not enough space to use a Metraloop.
Application and Installation
We have found that the construction requirements to install a functional seismic protection system for piping systems is ahead of the building codes themselves. Outside of any local codes, the only national code that we are aware of that gives us guidance of how to install a seismic device is NFPA 13. Although this is only mandatory for fire sprinkler water, it is a good basis of design for other systems if local codes do not exist.
NFPA 13 specifies that that the seismic device / flexible pipe loop be no more than 2 feet from the seismic separation. It also specifies that the bracing can be no farther than 6 feet from the seismic separation.
One of the most common seismic questions is; how do you fit in a Metraloop where the seismic separation is located between two walls that are close together?
You would need to create a seismic clearance in one wall as shown below. The seismic clearance needs to be the seismic movement plus 2 inches per NFPA 13. If you do not have this clearance a seismic event would shear off the pipe.
You would then place the Metraloop on the same side with the seismic clearance and put the brace on the other side of the Metraloop. If the customer is concerned with an open hole, this can be sealed with a sheet metal escutcheon plate that would yield during a seismic event and need to be replaced afterward. There are also companies that offer a rubber boot for these applications. Be sure to inspect all loops after a seismic event.
Another common question is if thermal and seismic movements can be combined into one devise. This is an important consideration since we have seen cases where the seismic movement had been addressed without any consideration of the thermal growth of the piping. One device can be used for both seismic and thermal movements can be done if:
1. The seismic device (Metraloop, V-Loop, or Dog Leg) is sized for the combined movement of the Thermal and Seismic loads.
2. The seismic braces next to the seismic device are replaced with guides. Typical Spider Guides with a corresponding pipe hanger is shown in the detail below.
Note: Make sure the guide is also sized for the combined movement.
In Line Seismic Joints
For installations where there is no room to install a Metraloop or V Loop, Metraflex has developed the Seismic Gator that is designed to handle movement in all directions.
The in-line profile is achieved by utilizing a combination of an externally pressurized joint, the Metragator, and a pair of gimbal joints on each end. The Seismic Gator is a rugged and reliable joint, but it will develop high thrust loads like an in-line bellows, unlike the Metraloop which develops very low loads.
Seismic Breakaway Hanger
Another Metraflex innovation that facilitates the successful installation of seismic joints in the toughest installations is the seismic Breakaway Hanger. The Seismic Breakaway Hanger is used as a support for a seismic device that will release in a seismic event allowing the seismic device to move freely.
See below details on how the Seismic Breakaway Hanger is installed:
To ensure your seismic devices (Metraloop, V Loop or Dog Leg) will perform properly in a seismic event, it is very important to use a Seismic Breakaway Hanger in the following situations.
1. Whenever the seismic device to installed to close to the deck of the structure. The seismic device needs to be installed at a distance of a minimum of two times the seismic movement minimum 12” or during a seismic event the seismic device will swing into the structure causing damage to both the structure and the seismic device. The picture below shows an example of a +-30 Dog Leg with a seismic Breakaway Hanger installed.
2. For applications with large movements that require longer hose lengths. The hose section will likely sag, so an intermediate support is required to prevent this sag. This is very important for plumbing system applications that require a proper pitch. The solution is a Seismic breakaway Hanger installed with a saddle on the hose as shown in the detail below.
3. To ensure that both sections of hose in a seismic device move freely and evenly. What a lot of designers and users do not understand is that in a seismic event very often one leg of a seismic loop takes all of the movement, while the other leg sees very little movement as shown in the details below. The simple solution is to install a seismic breakaway hanger in the support rod, so that in a seismic event it will release allowing both sections to move freely and evenly.
After any seismic event, the Seismic Breakaway Hanger can be simply reset.
How Not to Install a Seismic Device.
In the below picture you will see a set of seismic Metraloops installed over a seismic separation.
If you look closely, you will notice that the top of the loops is supported with a rigid hanger. If the structure on either side of the seismic separation moves in the Z axis (vertically) the sections of hose will either be pulled apart of over compressed. It is always important for a successful installation, for both seismic and thermal applications that the top of the loop be free to move freely.
In this case the solution is simply installing a Seismic Breakaway Hanger at the top of the loop. In a seismic event it will release, allowing the loops to move freely without damage.
Base Isolation Systems
Base Isolation seismic systems is an alternative to the traditional design with seismic separations. In a base isolation system, the entire structure is supported by base isolators that will allow large amounts of horizontal movement. Horizontal movements of 36” are common, vertical movements are usually 1 or 2 inches. This is an ideal way to retrofit seismic protection into an existing building.
Metraflex considers seismic joints for base isolation systems to be engineered products. Contact Metraflex or your local representative for more information.
The engineering staff at Metraflex has decades of combined experience.
Please feel free to call us to review your application.
Download white paper >
By James C. Neville Project Engineer, Engineering Services
Blue Ridge Numerics, Inc.
October 31, 2006
Conclusion: The CRV Flex provides a better flow profile entering the pump, therefore better pump performance, than a suction diffuser. The CRV Flex also has a significantly lower pressure drop.
Two CFD simulations are contained herein. Both analyze leading products used to condition and improve flow entering a pump. The first CFD analysis is of two popular styles of suction diffusers: the traditional cylindrical screen diffuser and the more recent model with a conical diffuser screen.
The second simulation is of the CRV Flex. A fixed vane device placed in front of the pump’s suction side elbow which imparts a half revolution to the media flowing through the elbow (see back cover). This minimal deflection of flow negates the turbulence caused by the geometry of the 90 degree elbow and produces measurably better flow conditions and
lower pressure drops than a suction diffuser.
Modeled were 4” x 3” reducing suction diffusers and a 4” x 3” reducing CRV Flex. Both designs were simulated at fluid velocities of 4 ft/sec and 10 ft/sec to determine overall pressure drops as well as the condition of the flow exiting the devices. Flow conditions between these ranges, slightly below and above can be interpolated from the data.
Simulations were conducted using CF Design version 9.0 from Blue Ridge Numerics, Inc.
A computational fluid dynamics analysis was performed on the Metraflex 4” x 3” suction diffuser to determine the overall pressure drop through the diffuser as well as the condition of the flow exiting the device. Furthermore, an analysis was performed
on a cylindrical screen design (constant screen cross-section) for a performance comparison. Both designs were simulated at fluid velocities of 4 and 10 ft/s. The simulations were conducted using CFdesign version 9.0 from Blue Ridge Numerics, Inc.
The CFdesign analysis setup is shown below in Figure 1. Additional pipe lengths were modeled upstream and downstream of the diffuser to ensure fully developed flow at the device and at the constant pressure outlet.
Various assumptions were made for the simulation of the suction diffuser and are listed below:
- Steady-state conditions
- Incompressible flow
- Water modeled at standard temperature and pressure
- Screen diffuser modeled as a distributed radial resistance (34% open area)
- Thermal effects negligible
Cut-surfaces showing pressure results are displayed in Figures 2-5 below. These cutsurfaces are oriented in such a way that they bisect the flow passing through the device. Figures 2 and 3 show results from the 4 ft/s inlet flow case while Figure 4 and 5 show those from the 10 ft/s case. The pressure drop across each design is summarized in Table 1.
Contours of fluid velocity magnitude through the device are shown in Figures 6-9 below. It is clear that several areas around the diffuser screen do not experience significant fluid flow and can be considered areas of stagnation.
The effect of the cross-like screen support can be seen in the fluid velocity results. Higher velocities were found in the top two chambers, especially in the conical diffuser analyses. It is shown that the cylindrical diffuser provides a more uniform fluid velocity among the four inner chambers.
Both designs showed that approximately 53% of the total fluid volume travels through the top two inner diffuser chambers, yet the difference in peak velocities between the upper and lower chambers was consistently higher in the conical design.
This shows that the fluid velocity distribution exiting the screen diffuser region is more uniform with a cylindrical screen diffuser.
A comparison of the flow profiles approximately 3 inches downstream of the device is shown below in Figure 10. In both the 4 and 10 ft/s scenarios, the cylindrical diffuser screen provided a significantly more uniform outlet flow.
Fluid particle traces released into the inlet stream are shown below in Figures 11-16. These traces help to visualize the fluid path as it travels through the device. In particular, Figures 13 and 16 highlight the areas of flow stagnation around the screen and the higher speed flow exiting the upper two inner diffuser chambers.
After reviewing the results obtained through CFD analysis, it is clear that the main difference in performance between the two diffuser screen designs is to be found in the downstream flow profiles. While the overall pressure drop through each suction diffuser is almost identical,
the downstream flow profile is more uniform with a cylindrical diffuser. Both designs showed similar areas of minimal fluid flow through several portions of the diffuser screen.
The Metraflex six-bladed CRV flow conditioner was analyzed to determine its effect in a 4” to 3” reducing elbow at water velocities of 4 and 10 feet per second. The simulations were conducted using CFdesign version 9.0 from Blue Ridge Numerics, Inc.
The CFdesign analysis setup is shown below in Figure 1. Additional pipe lengths were modeled upstream and downstream of the elbow to ensure fully developed flow at the CRV and at the constant pressure outlet.
Various assumptions were made for the simulation of the elbow and are listed below:
- Steady-state conditions
- Incompressible flow
- Water modeled at standard temperature and pressure
- Constant water properties
- Thermal effects negligible
Cut-surfaces showing the velocity profile and velocity vectors for both analyses are shown below in Figures 2 and 3, respectively. Note that there is very little discernible flow separation around the reducing elbow with the CRV in place. The CRV aides in providing a more uniform velocity profile beyond the reducing elbow.
A cut-surface of velocity approximately three inches downstream of the elbow is shown in Figure 4. Both analyses show similar velocity profiles, with slightly lower velocity regions near the bottom of the pipe due to the swirling action of the CRV.
The pressure gradient for both analyses is shown below in Figure 5. The total pressure drop through the CRV and elbow was found to be 0.37 psig for the 4 ft/s case and 2.19 psig for the 10 ft/s case.
Figures 6 and 7 below show particle traces released from various points on the pipe inlet. These traces show where individual fluid particles will travel as they pass through the system. Note the similar flow patterns shown for both flow rates. The swirling effect of the CRV is clearly visible.
The Metraflex CRV flow conditioner is shown to provide a near uniform velocity distribution downstream of the elbow. The CRV is effective in eliminating the large recirculation regions that would develop downstream of the elbow without a flow conditioner.
Standard CRV® Flex Configurations
A Comparison of traditional Y Strainers and Basket Strainers to the new LPD Strainer
Considering the y-strainer as a critical, energy-saving piping system component
Pumps, chillers, boilers and heat exchangers are the high-profile piping system components targeted to improve energy efficiency. More obvious in their energy consumption, these units are constantly being re-engineered, tested, measured and introduced to the market with improvements that measurably improve energy savings.
Other low-cost components in the system are routinely plugged into a design, accepted for their role in creating pressure drops and turbulence since there are no alternatives. One of these components is the strainer.
Today’s traditional strainer design was actually introduced over 100 years ago. The strainer was created to protect expensive equipment, such as pumps, chillers, boilers and heat exchangers by removing debris from fluid before it reached these components. But, a century ago energy consumption was not a consideration and energy efficiency was an unknown concept.
It is now better understood how strainers add pressure top due to their antiquated design and a screen that accumulates debris and increasing pressure loss. This pressure loss translates into pumps increasing energy consumption. It’s time to consider the strainer’s energy consumption.
A new strainer has recently been introduced. The new LPD y-strainer is redesigned from the ground up to provide superior pressure drop. Its internal geometry has been completely re-engineered, which also allows for a much larger screen with significantly increased debris capacity.
Following are comparisons between the centuryold y-strainer and basket strainer designs and the new LPD y-strainer to assist in making a more informed decision on which component is right for your piping system.
Y Strainers are widely used in hydronic systems and have been the most common type of strainer because
1) Unlike basket strainers they can be installed in horizontal and vertical pipe runs.
2) They can be blown down.
3) They are less expensive than basket strainers.
Basket strainers are the go to selection for strainers in “open” systems such as cooling towers. There were several perceived reasons for this.
1) Basket strainers have a higher capacity over the traditional “Y” strainer.
2) Basket strainers have a lower pressure drop.
3) Basket strainers are available with clamp on covers making cleaning the screen easier, then it would with a traditional “Y” strainer.
Taking a closer look at these points we will address some long-standing misconceptions about strainers, and discuss alternatives that will improve piping design.
The redesign of the strainer started with the basic Y strainer to take advantage of the versatility of vertical and horizontal installation, the ability to be blown down and the lower cost.
A CFD analysis and quickly confirmed the obvious, that the bridgewall universally used in Y strainer design caused restriction and an abrupt change in direction of flow, a big contributer to pressure drop.
In the new design, the bridge wall has been eliminated. See CFD analysis results below.
The bridge wall restricted flow and created turbulence in the strainer resulting in increased pressure loss (Figures 1 and 2).
The next design improvement – the pitch of the screen. Screen pitch was lowered from 45º to 22.5º.
Figure 3A: The old, traditional strainer, left, has a steeper screen pitch. Right, the new LPD Y-strainer has a shallower pitch allowing the screen to be much longer.
The lower pitch resulted in a much larger area of the screen to be in the flow path. This resulted in lower internal velocity and lower pressure drop (Figures 3A and 3B).
The change in pitch now allowed the screen to be much longer. Then as part of the redesign the casting was made larger to increase screen diameter (Figures 4A and B).
The result was not only a screen with much larger area for debris capacity, but a much larger screen area in the fluid flow.
Figure 4: Comparisons between traditional strainer screens and the new LPD Y-strainer
|Size||Basket Strainer Screen Area in²||Y-Strainer Screen Area in²||LPD Y-Strainer Screen Area in²|
Figure 4B: Four-inch strainer screens. Basket strainer screen, left, y-strainer screen, center, and LPD Y-strainer screen, right.
The results of all these design improvements result in a major improvement in the CV values of the LPD strainer, in some sizes almost double that for the old fashioned Y strainer and substantially better than a basket strainer as shown in (Figure 6).
Note, the CV values shown (Figure 6) are results from independent testing. We have found several basket strainer manufacturers literature that had much higher values that we believe to be overstated.
Probably the best feature of the LPD is its ability to capture debris. The same amount of debris in an LPD will have far less of an impact than it would in a traditional Y strainer or basket strainer. Using a dollar bill to represent the same debris blockage in each screen (Figure 7), it is obvious how much more unobstructed screen area is available in a 4-inch LPD vs a traditional, old-style 4-inch strainer.
Figure 5: Cv comparisons between traditional strainer screens and the new LPD Y-strainer
|Size||Cv Basket Strainer||Cv Y-Strainer||Cv LPD Y-Strainer|
Figure 6: Independent testing results comparing a clean LPD Y-strainer, and various amount of debris blockage, to a completely clean competitor strainer.
To demonstrate how the difference in pressure drop is even more dramatic between the two LPD and the traditional y-strainer, a test compared an LPD strainer with 92.7 in2 of blockage (29% blocked) still has a better CV than a 100% clean competitive strainer.
The larger capacity of the LPD strainer screen also means that you do not have to blow it down as often as a standard Y strain reducing maintenance costs.
Y-strainer vs Basket strainers
Basket strainers are considered easier to clean than y strainers. If shut off valves are properly located, the basket strainer does not lose water or require the system be drained to clean.
However, the LPD can be blown down where the basket strainer cannot. If blown down, the LPD may not even need to be cleaned.
The basket strainer is available with a clamp-on cover requiring only one bolt to be loosened to remove the cover. Depending on the manufacturer, clamp-on covers result in a lower pressure rating than a standard flange cover. This is normally a big advantage over a traditional Y strainer using a standard pipe flange for a cover. However, the new LPD design features a cover with 4 or 6 bolts depending on the size. Additionally, the cover of LPD strainers 10” and larger are hinged for easier handling.
Whenever a cover is removed from any of the traditional basket strainers or y strainers there is often a custom gasket that needs to be replaced. The LPD strainer uses a reusable O-ring seal.
In conclusion, it is taken for granted that science and technology has improved virtually everything around us. However, it has been over 100 years since technology has tackled the inherent flaws in y-strainer design. With the introduction of the new LPD y-strainer, significant energy savings have been captured with its new design. The LPD y-strainer is a significant improvement over other strainers, making it the best performing strainer on the market today.
Centrifugal pumps require well-developed inlet flow. To have a well-developed flow pattern, pump manufacturers’ manuals recommend about 10 diameters of straight pipe upstream of the pump inlet flange. Unfortunately, piping designers and plant personnel must contend with space and equipment layout constraints and can’t always comply with this recommendation.
Instead, it is common to use an elbow close-coupled to the pump suction which creates a poorly-developed flow pattern.
Poorly-developed flow can cause cavitation*, excessive vibration and/or poor pump performance.
The results can be premature failure of the bearing, seal or impeller, high maintenance costs and high power consumption.
Additionally, today’s pumps are not as tolerant to off-design inlet conditions as in the past.
Both flow velocity and impeller speed have increased. The result is a lower equipment cost for a given operating point, but the tradeoff is a pump that is more sensitive to inlet conditions.
For example, consider a
double-suction pump with a close-coupled elbow at the inlet. This type of pump is designed for flow to be split evenly between each side the impeller. The elbow creates turbulence that splits
the flow unevenly across the impeller (Figure 1a). The pump performance can be degraded, and the flow can create an axial imbalance on the shaft which would result in shortened seal, bearing and impeller life.
Cavitation is the formation of and subsequent violent collapse of vapor bubbles in liquid. Bubbles or cavities of vapor form in the liquid when static pressure drops below vapor pressure. These bubbles implode when the static pressure rises above vapor pressure, creating a highpressure shock. This implosion can be extremely damaging if it is in the vicinity of a pump component, and creates a distinct noise.
Elbows close-coupled to pumps cause cavitation in two ways: First,
as liquid passes through an elbow it experiences a drop in pressure
on the inside radius. Vapor bubbles form in this area and can be carried into the impeller. Second, non-uniformity of the velocity exiting the elbow can cause local acceleration at the impeller vanes, resulting in low-pressure areas and cavitation.
For example, consider a double-suction pump with a close-coupled elbow at the inlet. This type of pump is designed for flow to be split evenly between each side of the impeller. The elbow creates turbulence that splits the flow unevenly across the impeller (Figure 1a). The pump performance can be degraded, and the flow can create an axial imbalance on the shaft which would result in shortened seal, bearing and impeller life.
The Suction Diffuser Vane solves this problem by imparting a rotation to the flow upstream of the elbow. Flow leaving the elbow has an even velocity distribution as it enters the pump impeller (Figure 1b). The pump will perform closer to the factory rating, cavitation and noise will decrease, and seal, bearing and impeller life will improve.
Case History 1
Consider the feed piping configuration of a 3550 rpm double suction pump utilizing an 8-inch Suction Diffuser Vane, shown in
Figure 2. Prior to installing the Suction Diffuser Vanes, operation had
been noisy and the pump required
an overhaul every four months. A vibration test of the pump prior to the Suction Diffuser Vane installation is shown in Figure 3(a), and just after Suction Diffuser Vane installation is shown
in Figure 3(b). The dramatic reduction in measured cavitation (6-14 range on the frequency scale) also resulted in a reduction in impeller imbalance (1 on frequency scale) which reduced noise, extended seal, bearing and impeller life, and increased head and flow. Prior to Suction Diffuser Flex installation,
three identical pumps in parallel operation were required to deliver the required flow rate.
After the Suction Diffuser Flex was installed on each pump, two pumps met the flow rate requirement and the third pump
serves as an installed spare.
Case History 2
Performance and maintenance improvements can be achieved
with overhung end-suction centrifugal pumps too. Figure 4 shows an end-suction centrifugal pump which required a complete rebuild every six months. After installing the Suction Diffuser Vane, the pump has been operating for 1-1/2 years without a rebuild.
Case History 3
Dramatic improvements have also been witnessed with magnetically-driven pumps, which are extremely sensitive to cavitation and impeller imbalance. In particular, flow in one installation increased 4-1/2% from 1086 gpm to 1135 gpm after the Suction Diffuser Vanes were installed. At the same time, the horsepower draw dropped 7% from 43 to 40 and the discharge head increased 8.6% from 81 to 88 ft. The key benefit, however, was the reduced vibration which eliminated a frequent repair cost of $5000 on the shaft and bearings as well as an occasional severe-failure repair
cost of $14,000 per occurrence.