Friction loss by Steve Childers

Here is a very useful article written my a good friend of mine, Steve Childers from America.

It’s a System – Designing for Friction Loss
By Steve Childers

Historically I have always designed pond circulation systems as gravity fed to filtration, fully flooded pump(s) when at all possible (installed below water level), with pressure returns to the pond. Having done the calculations many years ago I have simply utilized an estimate of 4 ft of head for direct returns and 9 to 11 ft of head should this system incorporate a bead filter. Since I became involved in a project utilizing multiple airlift systems, I decided I better do some accurate calculations and thus put together an Excel spreadsheet to help with the process. If you care to follow along with this article, utilizing this spreadsheet, it is downloadable for subscribers at the KOI USA website in the archive section. Most of the following figures are from the downloadable friction loss calculator.

The first step in actually layout a pond system design is to know the designed volume of the pond and the desired turnover rate through bio filtration. For the sake of this article, we will work with a 10,000 gallon pond with two bottom aerated bottom drains at 6ft of depth with a bowled bottom. This is roughly a 12 by 24 ft deep rounded corners pond. The next step is to set our turnover rate parameter at 1 to 1.5 hours. Thus, each bottom drain will have to feed from 3750 to 5000 gallons per hour (gph). One constraint that needs to be considered is the “draw down” affect on the supply side of the pump and filter system for a gravity fed filtration system.  As we know, water will seek its own level. As water is draw from the filter at a given rate it creates a difference in water elevation between the pond and the gravity fed filter. The difference in water levels is determined by gravity and the pipe size for a given flow rate. In other words, the difference in elevation equalizes at the point there is enough gravity (weight of the water difference) from the point of origin (the pond) to the point of destination (the first filter chamber) to push an equal amount of water that the pump is pulling from the filter. This elevation difference will be multiplied by the number of gravity transfers. In other words, we know utilizing Bernoulli’s principles that a 4 inch line at a flow rate of 3780 gph will result in a draw down of  1 inch PER TRANSFER. Most gravity fed filtration systems will have at least 2 gravity transfers, Pond to Mechanical and Mechanical to Bio.

These gravity systems may have constraints as it relates to exit port elevations so understanding this concept is very important so that an oversized pump doesn’t draw more water than the system can supply. Thus, I tend to design for a 1 inch draw down per transfer and thus design at 3700 gph per drain.  This is right at our 1.5 hour turnover rate previously mentioned.

So, with our designed volume and turnover rates known, it is time to size a pump for this application. But first, although pumps may have a “rated” output, this output varies depending on the operating systems “head pressure.” In simple terms, the head pressure is a combination of piping friction loss and elevation (above the pond’s water level) that the pump has to pump to. Measuring the elevation is easy such as the vertical height of a waterfall. That being said, measuring the friction loss (in terms of ft of head) is what typically becomes a challenge for people, let alone then matching the desired flow rate and head pressures to a pump.

Utilizing data from the Plastic Pipe Institute for friction loss in schedule 40 PVC (fig 1) we see how many “ft of head” friction loss there is for varying sizes of pipe for each 10 ft of pipe.

Additionally, we need to consider the friction loss for the fittings between the pump and the discharge point. This is noted in equivalent ft of piping of the same size. See Fig. 2.

All of this information is readily available via the Internet in various forms and places including the KHA section of the AKCA website (www.akca.org). But, the real trick becomes equating that info to the actual design with the total number of feet and size of piping to be used along with the associated fittings. In addition, since a pump has a variable output depending on the operating head pressure and the friction loss is variable based on the flow rate (Fig 1), where do we go from here? Confused yet? If we utilize the rough design schematic in Fig. 3, we can estimate distances and fittings of each of the two systems. With the red piping denoting 4 inch lines and the blue denoting 2 inch lines. We can estimate the following in the design:

Length of pipe run approximately 10 ft from pump to side of pond (horizontal run in the diagram) plus the distance from the pump to the return of about 6 ft. (vertical on the diagram). These may be shortened in application by making more direct runs of the piping. In addition, the depth difference of the piping run also needs to be added in. If the pump is at the water level (but fully flooded) and the return height is 4 ft in depth, another 4 ft needs to be added to the piping. Thus, as drawn we have approximately 20 ft of pipe.

Now for the connections. Note that based on the data in Fig. 2, we want to use “sweeps” whenever possible/ Thus, we have the following “planned” fittings:

90 sweep exiting pump to horizontal. This can be eliminated if the pump head is switch so that the exit port is already at the horizontal.

90 sweep to the down vertical to desired depth.

90 sweep to the horizontal at depth

90 sweep towards the pond

90 sweep to TPR (return point)

Couple (acting as fitting for TPR)

Now is the point in time where we can actually calculate the total system’s pipe friction loss (in ft of head).

By plugging in our designed pipe length and fittings we see how pipe size affects the head pressures (in ft of head). See Fig 4. So are we done yet? Not just yet. Not all fittings are listed but “guestimates” can be made. As an example, I prefer to add a three-way valve on the discharge side of the pipe. To compensate for this, I would also add in a gate valve and also a “T” preferably installed with the “flow through” (verses utilizing a branch flow pattern). Although the Excel files is designed to act with input for  single line (size of  pipe) as in Fig 5, it can be used to get comparisons (such as Fig 4) or even combinations of different pipe and fittings on the same system such as in Fig 6.

Now, if when the design calculations are completed, more flow is necessary, increasing pipe/fitting size may be enough to reduce the head pressure enough to get to the desired flow rate. Thus, let’s say that we take and change piping size to 3 inch for the long pipe runs. If we change out 15 of the 20 ft to 3 inch and three of the 90 degree sweeps from 2 inch to 3 inch we get the results in Fig 6.

This results in the reduction of operating head pressure from the equivalent 42.8 ft of 2 inch pipe to 18.8 ft of 2 inch pipe plus 22 ft of 3 inch pipe. So what does that actually mean? What are the actual operating head pressures associated with Fig 5 and Fig 6?

Keep in mind that head pressures are dependent on the “flow rate”. Thus, for Fig 5 we have the results in Fig 7 (note, since our returns are at or below water level, there is no additional discharge height).

See Fig 8 for the resulting flows for Fig 6.

Now for the “Catch 22,” so to speak. A pump’s flow rate is dependent on operating head pressure and operating head pressure is dependent on flow rates. So where do we go from here? How do we “match” a pump to a designed system?

The first step is to plot the head pressure curve of the designed system. For the system in Fig 5 we have the curve pictured in Fig 9.

Now, if we have the performance curve(s) for pumps being considered, we can plot its associated curve overlaying our system curve in Fig 9. As an example we have in Fig 10 the pump performance curves courtesy WLIM Products for three specific pumps. Pump performance curves vary by manufacturer and by the specific model of pump. In other words, a one-quarter hp pump from one manufacturer is not the same as for another.

If we use the estimated data points from the one- eighth  hp performance curve we have the results in Fig 11.

Now these data points can be plotted against the friction curve for the result in Fig 12.

We have now solved the “Catch 22”. The point at which the two curves interest becomes the systems operating flow and pressure. In this example this become approximately 4000 gph at approximately 3 ft of head. Now, this is 300 gph more than our original design specification of 3700 gph. Will our supply side and filter system constraints accommodate that 300 extra gph? Instead of 1” of draw down per transfer (2” for two transfers), we will now have a little less than 1.1 inches of draw down per transfer utilizing the data in  Fig 13.

With less than 1.1 inches of draw down per transfer we are well within the supply side of the pump’s system requirements. The added 300 gph also allows for some ‘fudge factor” since actual installation rarely is as designed. Remember in the beginning of the article that I typically design for 4 ft of head? This design came in at 3 ft of head for 3700 gph.. What would a 4 ft of head do? If we simply add a 1 ft fudge factor into the additional Ht of discharge, we can see the results in figures 14 and 15.

Figure 15 shows that at 4 ft of head, the one-eighth hp pump produces just over 3700 gph and would be a well matched pump for this system design.

Now, the same exercise can be done for increases in the pipe size to reduce friction loss and increase flow such as what we specified in Fig 6. This results in Fig 18 and 19 respectively.

This shows how the system head pressure was reduced from 3 ft of head (without the added “fudge factor”) down to 1.7 ft of head on the designed system with the same pump. This results in an additional 200 gph from the 4000 gph up to 4200 gph. Since in this example the added 200 gph is not needed, it hardly justifies the added expense of the larger pipe size. This is not to say that for more remote filtration location (verses pond side as drawn) and thus longer pipe runs that this type of option may in fact become necessary to get back to the 3700+ gph desired.

Now, for systems that incorporate additional items such as a UV, simply estimate the like components in piping listed. As an example, a UV light with side inlet and outlet, add in 2, 90 degree elbows and the equivalent length of piping. Bead filter multi-port valves have a maximum pressure loss for a given flow rate. I typically use this Pound per Square Inch (PSI) loss even at lower flow rates. This gets us in the ballpark and adds a “fudge factor” since there can also be added loss from the internal piping and filter media. To convert PSI to ft of head use this formula:

PSI X 2.31 = Ft of head.

This can then be plugged into the calculator by adding it into the Additional Height of Discharge cell (Fig 21). If I look at my own gravity fed skimmer system that has a bead filter I have fittings listed in Fig 20. The multi-port valve on my bead filter has a max pressure loss of 4.8 PSI at 80 gpm. Utilizing the previous formula this equates to 4.8 x 2.31 = 11 ft of head.

Figure 22 depicts the system with the same one-eighth hp pump showing that we would have a flow rate of 1800 gph at just over 12 ft of head. This is not the desired flow and even substituting 3 inch piping where possible does not yield enough friction loss reduction to increase flow to a desired level. Thus, a different pump should be (and in actuality) used. If we substitute the data points for the one-quarter hp pump  from Fig 10 we get the results depicted in Figures 23 and 24.

By changing pumps to the one-quarter hp, we have increased the system flow rate to 2900 gph. This is an 1100 gph increase from the one-eighth hp pump. Granted, my example plumbing application was not what I would consider ideal with all of the added 90 degree sweeps and long pipe runs but sometimes we are faced with challenges that necessitate such. In the grand scope, it is the bead filter itself that contributes greatly to the higher operating head pressure of this system.

Of course the “as built” operating pressures can be checked with a pressure gauge installed on a straight section of pipe as close as possible following the pump but with at least 1 ft of straight piping before and after the gauge.

As I mentioned early on in this article, I put this together since I have become involved in an airlift powered pond project. We didn’t discuss this aspect and it will become a separate article in and of itself as that project progresses.

In designing a pond, it is important to match the pump to the system’s piping/operating pressures. Too big of a pump and the system can fail and too small of a pump can result in inadequate flows for the pond’s overall design and thus the pond as a whole could fail as well. It is also important since we want the most efficient pump in energy consumption for the water being pumped and at the desired flow rate. Although the process seems complicated, it is hoped that this article has served to provide a better understanding of the process and simplified such with the downloadable calculator for KOI USA subscribers (in the archive section at www.koiusa.com).

Click here to download the friction loss calculator