GCI TECH NOTES ©

Volume 3, Number 4

A Gossman Consulting, Inc. Publication

April 1997

One of the recurring problems we see in providing assistance to our clients concerns the operation of hazardous waste liquids handling systems. Common complaints center around flow problems and problems with pumps. Some of these
problems are associated with the failure of the equipment; pump seals that fail prematurely, flow control systems that don't control flow very well, fuel torches that self-destruct, etc. Some of the problems, however, result from a lack
of understanding of pump hydraulics and fluid dynamics principles as they apply to that most unique of fluids "liquid hazardous waste." It is not that liquid hazardous waste is an exotic one-of-a-kind liquid, it is more that it is
different from day to day. You can imagine an engineer contacting a pump manufacturer, a control valve supplier or an instrument vendor and attempting to answer their questions about the process conditions: "The specific gravity is 0.7
to 1.12, the viscosity may be 1 cps to 300cps, temperature is ambient except for the trucks that come in with hot tars from the refinery, flow rate is 150 gpm but only 15 gpm goes to the kiln abrasive? Well - it can be" What you end up
with is a liquids handling system that is just great most of the time, but has a few problems now and again. This paper can't possibly answer all of the questions that can be asked. Instead, we have attempted to remind you of a few of
the physical principles that make these systems work.

__Pressure and Flow Rate in Centrifugal Pumps__

First, a review of the relationship of pressure and flow rate in centrifugal pumps may be helpful. A centrifugal pump's pressure relationship and deliverable flow rate is represented by the specific pump performance curve. Each point on this curve represents a flow rate at a pressure. Pumping fluid through a piping system requires that the pump produce a pressure greater than that of the pressure differential losses (line losses) generated by the fluid flow through the system. As the pressure decreases, the flow rate will increase; and as the pressure increases, the flow rate will decrease. If the pressure is increased enough, the flow from the pump will stop. Typically, these pump curves are fairly flat at the higher pressures so the flow rate will not decrease at the same rate as the increase in pressure. As far as the pump is concerned, pressure is the independent variable and the flow rate is the dependent variable. However, within the piping system the line losses are directly proportional to the flow rate of the fluid. That is, the higher the flow rate the higher the line losses, the more pressure the pump must produce to maintain a specified flow rate. Consequently, within the piping system flow rate is the independent variable and the line losses (the back pressure produced) is the dependent variable. These line losses increase directly with flow rate and decreasing pipe size.

__The Effect of Increasing Viscosity on Flow Rate__

Variations in fluid viscosity, however, add an additional dimension to the problem. As the fluid viscosity increases, the pump discharge pressure needed to force the fluid at a specified flow rate through a pipe will increase for non-turbulent conditions (Reynolds Number <2000) and, theoretically, remain the same for turbulent conditions (Reynolds Number >2000). (This holds true for Newtonian fluids only, a discussion of non-Newtonian fluids is included below.) Unfortunately, centrifugal pumps are not capable of producing the pressures required at the higher and higher flow rates needed to achieve turbulent flow at viscosities above approximately 100 cps. Consequently, as the increasing viscosity causes the Reynolds Number to fall below 2000, at a given flow rate, the line losses increase dramatically in proportion to the viscosity. As noted above, the pump can only produce the flow rate and pressure as denoted by the pump performance curve. Once the line losses exceed the pump's rated pressure for that flow rate that fluid flow rate can not be sustained. When calculating process flows and pressures for a piping system, the designer will calculate a line loss versus a specified flow rate. If the line losses exceed the pump's rated pressure capability at that flow rate, a different flow rate is selected and a revised line loss calculated and again compared to the pressure/flow rate pump performance curve. In the end, the designer must balance the line losses (by selecting the piping diameter, limiting the viscosity range and setting a minimum and possibly a maximum flow rate) versus the capabilities of the pump. This frequently results in a pump and piping system that is over designed for any condition that is less demanding than the "worst case" condition for which it was designed. That is, the pump may be capable of a much higher pressure and/or flow rate than is needed for a particular fluid. (As an example, a fluid having a much lower viscosity than the viscosity for which the system was designed.) In actual practice, the pressure at any point in the piping system can only equal the back pressure created by the line losses of the piping and equipment downstream of that point. What happens then is that the pump wants to produce the flow rate corresponding to that pressure on the pump performance curve. If this is the case for a recirculation line (i.e. from a tank through the pump back to the tank) or for an inter-tank transfer line, the result may be the pump trying to run "out to the end of the curve", meaning that it is demanding more power from the motor than it is rated for which would result in the safety breaker tripping out on high amperage. If you have experienced this during a transfer, or while recirculating a low viscosity liquids, this was probably the reason. With a properly set safety breaker to protect the motor, this will be more an annoyance than any thing else until the operator learns the minimum pump discharge pressure required to preclude this trip-out. This discharge pressure can be attained by partially closing the pump discharge valve or by installing a back pressure control valve in the recirculation or transfer line.

__The Effect of Increased Specific Gravity of the Fluid __

Centrifugal pumps generate higher discharge pressures with increases in fluid specific gravity. Conversely, the higher specific gravity will increase the "head" (pressure) requirement the pump must generate to push the fluid to a
required elevation above the pump. This may all happen within the capabilities of the pump and you may never notice that the fluid specific gravity has increased. Alternately, the increased specific gravity may cause the pump motor to
be overloaded and trip out the same as discussed above for unrestricted flow rates even though the flow rate has not increased over that normally observed. The power requirement of a pump is proportional to the __mass__ of the fluid
pumped per unit time not just the flow rate. To transfer a fluid with a higher than normal specific gravity may require increasing the pump discharge pressure, as described above, by partially closing a pump discharge valve or a back
pressure control valve.

__Hot Days, Cool Days and Non-Newtonian Fluids__

Another problem that plagues HWF pumping systems is pump cavitation. Cavitation occurs when bubbles of vapor form in the fluid in the eye of the impeller and subsequently collapse as the fluid is forced out into the outer ring of the pump volute case. This is the result of the fluid having a vapor pressure sufficient enough at the pump operating temperature for these vapor bubbles to form at the slightly negative pressure in the impeller eye and then being forced back to liquid at the increased pressure as the fluid is "thrown outward" by the impeller into the pump discharge area of the pump volute case. The pump will produce a distinctive sound, often described as "pumping gravel". If ignored, the cavitation will destroy the impeller eventually giving it the appearance of Swiss cheese. Often, the cause of this cavitation is no more than that the fluid is warm because the day was warm, or the fluid is warm because the pump has been recirculating for a while and heated up, or the fluid has a higher proportion of material that has a lower vapor pressure than the last time it was received. In any case, this condition is easily solved by increasing the pump discharge pressure as noted above by partially closing the pump discharge valve or a backpressure control valve.

Non-Newtonian fluids are perhaps the most difficult pumping problem to solve. Non-Newtonian fluids do not behave the same as Newtonian fluids. A technical definition of Newtonian and non-Newtonian fluids goes something like this: "A Newtonian fluid has a linear relationship between the magnitude of the applied shear stress and the resulting rate of deformation. A non-Newtonian fluid would exhibit a non-linear relationship. "What this means is, a non-Newtonian fluid at rest will require more energy to get it to initially move and then less energy to keep it moving. It may even be the case that at higher flow rates a non-Newtonian fluid requires less energy to pump than a Newtonian fluid of the same "viscosity". It is here that we run into trouble. Viscosity can be measured in a number of ways. One method of viscosity measurement utilizes a cup with a specified diameter hole in the bottom of it. The cup is filled while first plugging the hole with a stopper, then the stopper is removed and the time, in seconds, it takes to drain the cup is measured. Another method is the familiar Brookfield viscometer, a set of spindles of a variety of diameters driven by a device that rotates at a number of selectable RPMs. This device measures the resistance to the torque applied to the spindle. It is possible to have a fluid that will not flow out of the cup and yet exhibit a fairly low viscosity on a Brookfield after the initial yield stress is exceeded. I have seen instances where "liquid" wastes would not drip out of an open hose but transferred from the tank once flow was established. We established the initial flow by pressurizing the tank until the pump began to pump, once started, the pressurization was not needed. I am

certain that every facility that burns HWF has received non-Newtonian fluids multiple numbers of times. In the vast majority of cases such receipt goes unremarked, but on occasion; "it's just that we can not get the stuff to move." What to do? First of all, does the stuff even pour? Next, take a look at what the Brookfield indicates at low RPM with a large spindle, does the value start out high and gradually get less? Compare the appearance to a fluid that pumps well that has the same "viscosity". This may not help you decide what to do to remedy the situation, but at least you will know that it is not the pump and the piping system. Sometimes all it takes is a little heat or a pump with a lower net positive suction head requirement, i.e. a slower speed centrifugal pump or a positive displacement pump such as a gear pump or a diaphragm pump. In any case, good luck.