Perspectives: Process Puzzler
Preclude Pump Problems
Achieving adequate performance requires multiple measures
THIS MONTH'S PUZZLER
Catastrophic pump failures at the tank farm are plaguing the commissioning of our new plant. The tank farm has an ordinary centrifugal pump (CP) as well as magnetic-drive CPs and a gear pump (see figure). The mag-drive pump started failing after the first few days. We got some warning from a pressure switch low (PSL) that flashed for a couple of minutes. We switched to a spare and it did the same thing after a day. The gear pump seemed to be operating fine, even after a few days — but a reading on an infrared gun indicated the casing was hot; the downstream flow meter showed a dropping flow rate. An operator loosened the packing on the pump while I was away at lunch and it now runs fine. The water pump is showing the same symptoms as the mag-drive pump but hasn't failed yet during the first week of commissioning. Are we out of the woods?
CAREFULLY CONSIDER TANK LEVEL
I have the following observations and comments. Regarding the first pump, assuming it is a metal magnetic-coupled pump: the flow rate may be too large, causing the pump to run beyond its operating range. This may be caused by the bypass if the orifice pressure drop is too small, or the flow rate to the consumer is too large. It might be advisable to check the flow rates, for example, using a clamp-on ultrasonic flow meter. At a high flow rate, a magnetic-coupled pump may encounter too high radial forces and a loss of lubrication in its sleeve bearings, causing the failure, before motor protection can cut in.
Now, consider the water pump. This pump also may be running at too high a flow rate, not yet triggering motor protection, since flow seems not to be restricted by a control valve, etc. This flow rate may also be checked using a clamp-on flow meter.
I assume that suction-side strainers have been checked for both pumps and they are not (partially) blocked. I also assume that PSLs for these pumps are set to pressures higher than the geodetic height, i.e., the full level of the tank over pump.
These failures may be aggravated by a high liquid level in the tanks, which reduces the pressure drop the pump has to overcome. At high level, the pumps may seem to be oversized. Using power consumption measurements to check operating points is another way to determine the pumps' flow rates, but this lacks precision. My advice is to use flow controls to restrict the pumps' operating ranges.
Dr. Walter Schicketanz, consultant
Ing. Pumpenfachingenieur GmbH, Rosenheim, Germany
LOOK AT THE SUCTION STRAINERS
If you read many installation manuals you will find that a medium strainer (40–60 U.S. mesh) often is used to catch debris during startup from suction lines. A tank farm application might consider a 100-mesh self-cleaning suction strainer; above 30 mesh the strainer basket must be reinforced. So, let's consider the details.
It sounds like you need to replace the elements in the strainer with coarser ones. Usually, this is a U.S. mesh 3–8 (with holes: 7.42 mm and 2.54 mm, respectively) but to be more exact you should refer to the pump cut sheet; I've seen 500 mesh used in pump suctions. In these cases, opt for a coarse strainer upstream and have your controls watch for flow restriction or pressure drop, say, by an increase in control valve position.
The pressure drop can be estimated by the following equation:
∆P = (Q/Cv)2×SG×C1×C2
where Q is flow rate in gal/min, Cv is a vendor-supplied flow value for the filter, SP is specific gravity, C1 is the mesh correction factor (C1 = 1 for 8 mesh and below) and C2 accounts for the % of area clogged. Vendors usually use C2 = 2 for 50% clogged. A rule-of-thumb, if you don't have any other data, is to take strainer pressure drop as 1 psi at the (maximum) design flow.
For the mesh, look for the maximum solids (assumed diameter) that the manufacturer allows. In Europe, suction strainers are specified as requiring a hole diameter of 67% of the pump manufacturer's maximum solids size and a surface area 3–5 times the pipe cross-sectional area. Don't run a pump without a strainer because this violates your warranty and could put you out of business; I've seen zebra mussels and even a big chunk of a pallet go through suctions and demolish impellers.
As for the operator, he pulled a fast one. The gear pump is operating without a basket. Discourage this practice by locking down the strainers as you would lock-open a valve. Gear pumps have even tighter constraints than centrifugal pumps: a gear pump won't survive a pallet. Consult the manufacturer for specific details with equipment like gear pumps and wet gas compressors; one compressor manufacturer listed 11 mm as the minimum impeller clearance but this assumes there is a working mesh pad in the upstream knockout drum — always consult the manufacturer. I guessed 67% of 11 mm for the compressor and later was proved correct.
Some pumps are even more delicate. An air diaphragm pump usually requires a 40–60 mesh strainer because of the piston check valve.
Now, let's consider some of the other reliability problems. First, you don't need double seals for a water pump. Single seals, packing or both in tandem will suffice. Because gear pumps are not pressure-variable, you should use a flow alarm to catch problems such as missing teeth and a blocked valve. Magnetic-drive and seal-less pumps pose a particular problem because of tight clearances. As a general rule, a PSL is not an effective tool even with double-seal centrifugal pumps used as the baseline in the refinery and chemical industry. Use a flow meter or switch, not a pressure switch, to detect a problem with a centrifugal pump.
You may want to look at your net positive suction head available (NPSHA) calculations. Strainers often are treated as pipe fittings even though they're not.
Dirk Willard, senior process engineer
Ambitech Engineering, Joliet, Ill.
OKLAHOMA STATE STUDENTS OFFER IDEAS
This puzzler was used by the Chemical Engineering Graduate Student Association at Oklahoma State University, Stillwater, Okla., with CP's approval, for a process troubleshooting contest. Graduate and undergraduate students participated in the contest as teams. Dr. Russ Rhinehart, Dr. A. J. Johannes and Dr. Rob Whiteley judged the entries. The event was partially sponsored by the Graduate and Professional Student Government Association (GPSGA) at Oklahoma State University.
JOINT FIRST PLACE #1
General assumptions on which proposed solution is based:
1. Power supply is not fluctuating.
2. Pumps have been sized and rated for selected duties.
3. Material of construction is compatible with service fluids.
Magnetic-drive pump servicing organic solvent with high vapor pressure. Magnetic-drive pumps are used when we want to prevent direct contact of the fluid and drive coupling and also to eliminate any possibility of leak. A PSL is activated when the pressure on the suction side of the pump falls below a minimum pressure. Commonly, the pressure drops below the minimum pressure if there is no feed at the suction side of the pump. For the suction side of the pump not to have any feed, the following events might have occurred:
• The organic solvent tank was completely drained off.
• The strainer was partially or completely choked, preventing any liquid from flowing into the pump.
We are assuming that the level transmitter (LT) in the organic solvent tank was calibrated and installed appropriately, and the low level and low-low level alarms in the distributed control system (DCS) were set right. Check the calibration documentation. The low level alarm ought to have displayed when the level dropped below the set limits. So, it will be worthwhile checking the level alarms on the DCS and correcting them if they were off. If the alarm settings are right on the DCS, then the likely cause is strainer blockage. In many cases, the reason for pump failure during commissioning activities is a blocked strainer due to debris from installation. Therefore, check the strainer and clear any debris.
A third but less probable event that could have occurred is generation of vapor bubbles when throttling the solvent back in the suction line through the restriction orifice. When a high-vapor-pressure solvent from the discharge side of the pump flows through a restriction orifice, there is a possibility of some of the liquid getting flashed. When this vapor mixes with the suction side of the pump, the possibility of some vapor going through the pump exists. Vapor will cause the pump to cavitate. Also, temperature will build up on continuous operation and as more organic solvent is recycled, due to churning of fluid in the pump casing. You may consider installing an intercooler before the restriction orifice to avoid flashing. We cannot rule out the combined effect of a blocked strainer, low level in the tank and cavitation — leading to the PSL alarm and failure of main pump. It also is possible that some solvent vapor existed in the suction line and, once the pump was switched over, led to cavitation in the spare pump as well. For smooth operation and maintenance, we recommend having two strainers in parallel in the common suction line, upstream of both the main and spare magnetic-drive pumps. We also recommend that you install pressure transmitters in the suction line with set points such that the minimum suction pressure always is maintained above the vapor pressure of solvent, i.e., to maintain net NPSHA above net positive suction head required (NPSHR). This will avoid any vapor from going through the pump.
Gear pump with packing serving viscous reactant. The dropping flow rate in the gear pump servicing the viscous reactant likely was because of excessive back slip. For a gear pump to deliver a constant flow rate, the viscosity, differential pressure and rotations per minute should remain constant. There is bound to be a small amount of back slip in all gear pumps because of the clearances meant for the pumping parts to move. However, to keep a constant delivered flow rate, this back slip should remain a constant. If the differential pressure between the inlet and outlet side increases, the amount of slip would have increased, thereby leading to a drop in flow rate. We suggest you check the differential pressure data during the time of commissioning to determine if differential pressure caused a drop in flow rate. If it did, the next question will be to identify what caused the change in differential pressure in the first place. To have a constant flow rate, we recommend installing a tachometer to measure motor speed and then tying [the reading] into a control loop that will increase or decrease the motor speed to keep the delivered flow rate constant. Alternatively, you can consider using a control valve to bypass excess flow rate back to the inlet side to keep the delivered flow rate a constant.
Centrifugal pump with double seal servicing water. As mentioned in the magnetic-drive-pump troubleshooting, a PSL is activated if the pressure in the discharge line of the centrifugal pump goes below a minimum pressure set point. The reasons that a PSL will activate on the water pump are similar to those for the magnetic drive pump — a blocked strainer or no feed on the suction side. The first step here would be to check and clean the strainer of any entrained debris. If there was no debris, then the tank might have been drained off, activating the PSL. However, when a tank drains off completely, the level alarms would have displayed on the DCS before the PSL. We are assuming that the level alarms in the DCS were programmed right. The order of the alarms should have been a series of low level alarms followed by the PSL. Taking the process and instrumentation drawing at face value, the LT is shown installed at the bottom of the tank and the high-pressure leg of the LT is below the suction line of the pump. This arrangement will cause the low level alarms to display only when the water level falls below the suction line of the pump. Consequently, the PSL will display before the low level alarms, thereby confounding you. Therefore, check and correct the installation of the LT. In general, an LT is installed at the top of the tank if it is a radar type LT. Again, for smooth operation and maintenance, we recommend two strainers in parallel in the common suction line, upstream of both the main and spare pumps.
Anand Govindarajan, PhD candidate
Roshan Patel, PhD student
JOINT FIRST PLACE #2
Tank 1 — Organic solvent with magnetic-drive centrifugal pump. The organic solvent present in the first tank is highly volatile, which would cause cavitation in the centrifugal pump. Cavitation occurs when fluid pressure is lower than the vapor pressure; this causes vapor bubbles to form and implode in the pump. Therefore, the fluid pressure must exceed the vapor pressure at all points in the pump. If cavitation exists, the pump will be less effective, yielding a lower-than-expected pressure downstream. Initially, the tank would have been filled with the organic solvent, which means there is large pressure head located at the bottom of the tank. As the level in the tank decreases below the initial height, the pressure head decreases, which in turn might cause the fluid pressure in the pump to decrease below the vapor pressure. When the pump is replaced with the spare and the tank is refilled to its original height, the pressure head is recovered to its original value. As soon as the spare pump is activated, the tank level starts to decrease again, resulting in a lower pressure head. Hence, given the current situation, it is highly likely that there is cavitation at the pump inlet due to insufficient pressure head. The level of the tank should be checked whenever the PSL flashes. The solution to this problem is to maintain the level in the tank to keep the pressure head sufficiently above the vapor pressure to prevent cavitation. This can be done by increasing the pressure in the tank with an inert gas blanket on top of the organic solvent. If the structure of the tank cannot handle this increase in pressure, a heat exchanger can be added before the pump to ensure the fluid does not vaporize. Another option is to place the pump at a lower elevation, which would cause the NPSH to increase. If the organic solvent or the operating temperature near the pump were known, one could determine if cavitation is truly the problem by comparing the pressure of the fluid to the vapor pressure of the solvent.
Tank 2 — Viscous reactant with gear pump packing. Packing failure in gear pumps typically occurs from over-tightening the packing material. The packing on the gear pump is loosened to allow minor leakage, which lubricates and cools the sealing surface, ensuring the seal has a longer life. The infrared gun detected overheating in the casing, which was caused by the packing being too tight. This did not allow sufficient fluid leakage for lubrication and cooling of the pump. Another reason for the casing to overheat may be due to the friction caused between the seal and the shaft as a result of over-tightening. Based on the present scenario, the pump is overheating due to insufficient lubrication as a result of over-tightening of the packing material. The solution is to ensure the packing is adjusted proportionally to the leak rate. As the pump heats up, the fluid becomes less viscous, which in turn causes slippage of the fluid in the gear pump. An increased amount of slip causes a decrease in flow rate downstream.
Tank 3 — Water with double-seals centrifugal pump. Double seals are utilized in a centrifugal pump for a fluid that operates at a high temperature or is toxic, abrasive or volatile. The primary reason to use a double seal is if water is used at a high temperature. This particular scenario would make sense if the tank farm were commissioned in the summer months. As the fluid temperature rises, the vapor pressure increases, which escalates the risk of cavitation. If cavitation occurred, there would be a lower pressure downstream of the pump. Another reason for cavitation would be if the original height in the tank dropped. This results in a corresponding decrease in the pressure head, which in turn would lower the fluid pressure below the vapor pressure of water. Compared to the organic solvents in Case 1, water has a lower vapor pressure, which yields a lower risk for cavitation. There is a high chance the pump is failing to perform due to cavitation as a result of insufficient pressure head and higher water temperature. To check for the cause, the performance of the pump must be observed with respect to the water level in the tank and water temperature. The solution is to maintain the level in the tank so that the fluid pressure remains above the vapor pressure. Another possible solution would be to pressurize the tank with an inert gas blanket such as in Case 1 if the tank can withstand this increase in pressure. Another option is to install a cooler upstream of the pump.
Leigh Krause, MS student
Kaston Murrell, MS student
Jagdeep T. Podichetty, PhD
The tank farm is not out of problems. Low suction-side pressure of the pumps connected to the tank is causing the problem.
Magnetic-drive pump issues. The choice of the magnetically driven centrifugal pump is logical considering a volatile organic solvent. This would avoid leakages and thus enhance safety over a motor-driven centrifugal pump. The important point to note here is that generally magnetically driven pumps have lower efficiency than motor-driven pumps.
The flashing of the PSL suggests cavitation taking place inside the magnetic-driven centrifugal pump. Cavitation usually occurs when the NPSHA is less than the NPSHR. The NPSHA for centrifugal pump is computed as the difference between the sum of static head and absolute pressure at the liquid's free surface and sum of suction line losses and vapor pressure of the liquid. In this case of the organic solvent tank with the magnetic-drive centrifugal pump, there exists a suction head and not a suction lift because the source of supply of the liquid is above the centerline of the pump. The suction head is represented as the vertical height between the free level of the fluid to be pumped and the centerline of the pump. The liquid's vapor head is high because of the high vapor pressure of the liquid in the tank. This is because the liquid's vapor head is directly related to its vapor pressure. This difference in the NPSHA and NPSHR can occur in various cases. One of the causes can be insufficient pressure head on the suction side because of insufficient height difference between the liquid level and centerline of the pump impeller.
Centrifugal pump issues. The water tank with the centrifugal pump also might have the same problem as the organic solvent tank with the magnetic centrifugal pump. In order to avoid this problem, the available NPSH can be increased. The tanks with the magnetic and ordinary centrifugal pumps might not have issues if the NPSHA is increased such that it exceeds the NPSHR.
Gear pump issues. Gear pumps generally are used to pump high-viscosity fluids. Cavitation in a gear pump usually is not as intensive as that in a centrifugal pump. Though there is drop in flow rate, the pump did not fail, but the pump casing became hot. The equivalent term for NPSH for any positive displacement pump is net inlet pressure requirement (NIPR). Similar to other pump cases, the low inlet pressure led to less NIPR than required. This increases the pressure drop across the gear pump. This pressure drop couldn't produce work because of the dropping flow rate. All the input energy started transforming into heat and, hence, the casing started to become hot. Loosening the packing reduced the pressure drop across the gear pump, making it work better.
Investigation results/suggestions. Thus, we narrow down to increasing the available NPSH and NIPR for the pumps. There are several strategies to achieve this:
1. Lower the pumps a few meters into the ground (data and calculations would give the actual distance the pumps need to be lowered.)
2. Frictional head losses in the pipe can become significant if the pump is located far away from the tank farm. This problem can be overcome by placing the pump as near to the tank as possible (considering safety issues).
3. If the first two steps do not suffice, then the following options can be considered:
• Increasing the height of the liquids in the tanks (increased solvent inventory);
• Pressurizing the solvent tank with an inert gas to generate additional head (additional equipment and inert gas supply); or
• Reducing the operating temperature (additional utility requirement).
Upasana Manimegalai Sridhar, PhD candidate
Amey Thorat, MS student
The warnings and indications, analysis and solution to the pump failure problems are provided below.
1. Cavitation of the magnetic-drive centrifugal pump for organic solvents with high vapor pressure:
Warnings/indications. Two warnings of cavitation in the magnetic-drive centrifugal pump for organic solvents are seen in the troubleshooting memo, while the third can be checked on.
• Flashing of the PSL light indicating low pressure at the pump;
• High-vapor-pressure fluid being pumped.
• Cavitation will be confirmed if intermittent discharge flow rate and pipe vibrations are seen.
Solution. Apply energy balance/Bernoulli's Equation between point one at the exit of the tank and point two at the inlet of the pump (noting that the pump is at ground level and assuming a negligible velocity change from point one to two). If Bernoulli's Equation is not satisfied, the solution to the problem is to raise the tank height, which will provide necessary head to avoid the cavitation.
2. Cavitation of the magnetic-drive centrifugal water pump:
Warnings/indications. The following warning sign of cavitation is seen in the memo, while the second indication can be checked for.
• Again, the PSL indicator flashed for this pump. The pumped fluid in this case is not a high-vapor-pressure fluid but we suspect that water flowing with high velocity through the pipe creates a vacuum at the eye of the impeller (assuming there is no vent to the pumping system).
• If intermittent/irregular water flow rate and pipe vibrations are observed, cavitation of this pump will be confirmed.
Solution. The same solution applied to the cavitation problem in the magnetic-drive centrifugal pump for organic solvents can be applied here.
3. Choking of the gear pump:
Warnings/indications. The warnings, indicating choking of the gear pump, are as follows:
• An infrared gun indicated the casing was hot.
• The downstream flow meter showed a dropping flow rate.
• The packing on the pump was loosened, which made the pump run fine.
Analysis. First, fluid begins to block the gaps between the teeth of the gears, dropping the flow rate downstream. Thus, the inlet fluid and discharged fluid flow rates are not equal. The fluid starts accumulating inside the casing, particularly by deposition on the casing walls. If the pump is run at this condition, less energy is used for pumping due to the lower flow rate. Part of the energy supplied is used for pumping while the remaining energy is dissipated as heat by means of high shear stress induced by high viscosity fluid. This makes the casing hot. The choked viscous fluid also creates an obstruction to the smooth rotation of gears. Loosening the packing solves the problem — and supports the previous statement. It may show some fluid leak through loosened packing, which will confirm choking of the pump.
Solution. Implementation of the following, based on their practicability, would help to eliminate choking of the gear pump:
• The pump casing should be opened and internals cleaned periodically to help avoid this problem.
• A viscosity-reducing agent should be added to the liquid if permissible and possible.
• The pump suction line should be heated using steam tracing and the fluid tank using a heating jacket.
• The gear pump should be replaced with a screw pump if the fluid viscosity cannot be reduced and the problem persists after the above changes.
Nikhil Japtiwale, MS
Carrie German, MS
Magnetic-drive failure scenarios. In the case of the magnetic-drive pump, we must explore four discreet scenarios by which the pump could be failing. The first possible scenario is the occurrence of cavitation due to insufficient heat removal, as a solvent with a high vapor pressure will start to flash at temperatures relevant to this particular pump. Additionally, the NPSH could be too low, either by design or by an incorrect reading from the corresponding LT. To eliminate concerns that are design related, the pump curve should be consulted to ensure that it matches the installation. We also must note that there is a minimum flow bypass line recycling flow to the line upstream of the mag-drive; typically this practice is used to maintain a minimum flow based on the pump's design, however, the bypass line should (most likely) spill back to the tank.
The gear pump. It exhibits an unusually high casing temperature, which is indicative of the packing not being installed correctly. We are told that the packing was loosened, and the gear pump began to operate as expected. The packing, in this case, was the culprit for friction-related heating.
The water pump. In this scenario, the LT is located at the bottom of the tank, so it may not be reading correctly, and if the NPSH is low, the pump would act as described. There are numerous reasons for the LT to be misreading, including a situation where the line to the LT is plugged, as well as being miscalibrated. The pump curve should be referenced and checked against the installation; this particular pump may require a minimum flow to function correctly. (If the flow is not achieved, the pump will heat up.) We also must note that double seals are not typical for water pumps; therefore, the installation and seal materials should be checked.
For all pumps. Something to consider for all pumps is that the impellers could be too large for the service of their respective liquids. If this is true, the solution would be to implement a minimum-flow bypass line back to the tank.
To answer the question: No, we are not "out of the woods" yet, and, in all cases, the pump curves should be consulted.
Daniel Montoto, sophomore
We manufacture product in a 4,000-gal reactor operated at 150 psig that has a conventional steam-heated jacket. We're considering a project to cool a batch after a reaction using an external plate heat exchanger: the project involves doubling the size of the product pump and adding a plate-and-frame exchanger. Presently, a shell-and-tube exchanger handles cooling: product flows through the shell side of a 1,2 exchanger (2 shell passes); 85–115°F cooling-tower water provides cooling. The area is about 2,500 ft2. Currently, the contents of the reactor take four hours to heat up with 35-psig steam, and fifteen hours to cool down with cooling tower water, before being pumped to storage as product. The reactor's jacket provides heat for initiating the reaction and heating the ingredients in the early steps of the batch. A waxy material added in the beginning would form a second phase and precipitate without heating; this problem disappears when the material reacts with the aqueous phase. A multi-speed agitator, running at 120 rpm early in the batch, often is turned off altogether because of foaming in the middle of the process cycle. The only circulation then is by a 100-gal/min pump circulating fluid through the product heat exchanger with water off. We want to cool the product from 250°F to 125°F, near the product gel point. Can you suggest a better way to speed up cycle time and reduce heating and cooling time?
Send us your comments, suggestions or solutions for this question by June 13, 2014. We'll include as many of them as possible in the July 2014 issue and all on ChemicalProcessing.com. Send visuals — a sketch is fine. E-mail us at ProcessPuzzler@putman.net or mail to Process Puzzler, Chemical Processing, 1501 E. Woodfield Rd., Suite 400N, Schaumburg, IL 60173. Fax: (630) 467-1120. Please include your name, title, location and company affiliation in the response.
And, of course, if you have a process problem you'd like to pose to our readers, send it along and we'll be pleased to consider it for publication.
More from this perspective...
Readers help sort out a sulfonation in shambles.
Readers suggest how to improve a blending operation.
Readers identify the cause of poor absorption in flue gas system in this month's process puzzler.
Readers identify problems with a relief system in April's Process Puzzler.
How would you improve this design?
Readers suggest possible causes for a mysterious mixer explosion in Chemical Processing's monthly Process Puzzler feature.
Readers suggest how to keep sludge from clogging equipment in this month's Process Puzzler.
Readers suggest ways to get a second-hand unit to perform better
Readers suggest how to preclude ignition-temperature-test failures.
Readers suggest ways to avoid temperatures swings and trips.
Readers raise a variety of issues to consider.
Readers decipher a compressor surge.
Readers offer tips on how to keep a standby compressor ready for service.
Readers suggest ways to maintain performance on hot days.
Readers offer ideas for improving reliability of a weigh feeder in Chemical Processing's monthly Process Puzzler feature.
Several steps can help handle gasoline-laden vent gas.
Several factors can prevent two sensors from giving matching readings.
Readers recommend significant upgrades to a relief system.
Chemical Processing's readers offer tips about how to properly isolate a waste heat boiler.