Margin of Safety vs. Safety Factors
There I was, working hard spotting and connecting dots with a steady hand and focused eye, when my boss came in and announced, “We are doing a water flow test this weekend and I want you to attend. I want you to learn how to do a flow test.” I was ecstatic. That’s right, I was like a kid at Christmas. I am sure many of you are questioning just what kind of guy gets excited about doing water flow tests, but you must understand, it was going to be my first one. Up until that point in my short eight-month career, I had heard, read, and seen pictures about them, but never had an opportunity to witness one live let alone participate in one. So, this was a big deal. Also, I learned how to do hydraulic calculations within a few months of starting my dot-spotting career and had been performing calculations for some time but the reality of 500 gpm or 50 psi… well, let’s just say you could have been talking to me about quantum physics and I really would not have known the difference. I was finally going to get to experience a full-blown water-supply test.
I remember that Saturday morning like it was yesterday. They had shut down parts of downtown so the street we were conducting the test on would be clear of vehicles and pedestrians. The side streets were lined with city water department trucks along with police cars and fire trucks. I can remember thinking that this was going to be very cool. After a short meeting with those of us who were actively involved, we took our positions at the static and residual hydrants to ready ourselves to flow some water. The anticipation was growing. And, as is the custom, the “new guy” got the assignment of helping on the flow hydrant, destined to get wet for sure. The “all clear” was given and we started opening the hydrant. The sound of the air evacuating the hydrant barrel was much like a countdown to something launching in the air. The pitch of the air got higher and higher as the rush of water began to fill the hydrant making its way to the open outlet we had set up to measure from.
Finally, water began to spray out, more with each turn of the hydrant wrench. The stream of water shot all the way to the other side of the street. At last I was up to bat and all eyes were on me as I stepped up to the hydrant to perform the pitot reading. Prior to the start, my boss had spent several minutes going over the finer points of doing this test – how to hold the pitot tube, how to find the “sweet” spot, how long we needed to try and hold it for a good reading – yes, he told me everything I needed to know except one thing. He forgot to tell me to “Hold on tight to the pitot tube”! Yep, as you can imagine, no sooner than I reached down to insert the tube into the stream but the tube went flying out of my hand and across the street slamming violently against the curb on the other side. The test was over. Everyone had to go home; the game was over. Steven ruined it for everyone.
You could say that was the day that I learned about pressure and flow. We rescheduled the test and just as monumental as the first attempt, the second attempt proved to be even more so because after we performed the test (and yes, he let me try again), my boss punched some numbers on his calculator and announced to me, “That was 1,100 gallons per minute.” It was at that exact moment that my career in fire protection started. Even though I had been performing hydraulic calculations for several months prior, making sure the dot was under the line, all those numbers meant to me were just that – numbers. That day I “saw” 1,100 gallons per minute and I “felt” what 43 psi was and it all made sense to me. The numbers became real.
The epiphany came a year or so later when doing a routine flow test near a golf course. It was early in the morning and the golf course lawn sprinklers were all running. The city water system was known for its fluctuations because of the golf courses, and so we all knew that the best time to do a flow test was early in the morning when the city water pumps were on. The conditions for the test were perfect; meaning, I could get myself situated comfortably at the outlet so that I could really do a good job of steadying my arms and hands to get a good reading. But, no matter how steady my hand was, the needle of the gauge was jumping up and down, typical to every flow test I had done up until that point and ever since. It was then that it hit me again. That is my degree of accuracy! Somewhere between 55 psi and 65 psi was the residual pressure of this water system. Here I was struggling over an accuracy of hundredths of a decimal point when the actual gallons per minute and residual pressure where based on a degree of accuracy of plus or minus 5 psi! The reality of “accuracy” was illuminated again the first few times I recalculated as-builts from a couple of projects and realized how quickly my perfect system, based on this bouncing needle was erased by a couple of HVAC ducts and a roof drain!
That is when I began to dissect the calculation process and just how close I needed to get that demand dot to that supply line curve (the 1.85 graph). The “degree of accuracy” became the bane of my career for several years after because it translated into a “floating node.” That, and of course the software that we all were using was reporting to the hundredth of a decimal point which gives the perception that we are dealing with a high degree of accuracy. Now, looking back it seems so silly, but then, it was serious because many of us “actual practitioners” had no clue where numbers like 1,500 ft2 or 0.10 gpm/ft2 or 130 ft2 came from. Most of us were handed a red book and told to get that dot as close to that water supply curve as possible, while others were told to get it as close as 10 percent of the curve and still others told 10 psi of the curve. In other words, we were given a margin of safety to achieve – not one pound under nor one pound over.
So, just how close should we be sizing these life safety systems? Don’t go looking in NFPA 13, Standard for the Installation of Sprinkler Systems. You won’t find it there. It is left up to us, the user. The hundredth of a decimal place? The nearest 10 psi? My answer is: it depends. I say that because many of us now know the numbers we use in the calculation process come with “built-in” safety factors. There are several places we can look to find degrees of accuracy relative to built-in safety factors; for example, the sprinkler K-factors are really an average passed on various amounts of water flow, and the coefficient of friction or C factor is not constant over time even though we treat it that way. For me, however, there are three safety factors that are most significant when considering how close to the curve my dot should be or in other words, safety margin.
Area of Operation
The first one is the area of operation, otherwise referred to as the remote area. With each hazard level, there is a prescriptive density to be delivered over a specific area (square feet). For instance, light hazard occupancies demand a minimum of 0.10 gpm/ft2 over a minimum area of operation of 1,500 ft2. Let’s bypass the discussion on where 1500 ft2 came from, but rather focus on the equation we use to draw out the 1500 ft2 on our design. As shown in the Annex for Chapter 23.4.4 of NFPA 13, we are instructed to take the 1500 ft2 and multiply it by 1.2. Ever wonder why? You realize that we are taking a square that is 38.73 ft on all four sides and multiplying one side by 20 percent, which means it’s not going to be a square anymore but rather a rectangle. That rectangle is going to be 20 percent longer on two sides than the other sides. So, again I ask why? If you keep reading we are told that we are to orientate this rectangle with the longer sides, which are now 46.48-ft long, parallel with the branchlines (38.73 x 1.2 = 46.48). We usually assume that a fire is going to grow equally in all directions or circular. However, we orientate the area of operation to simulate the worst way a fire could grow in terms of hydraulics. That is, the worst way a fire could grow and cause the greatest hydraulic demand would be parallel with a branchline. This would cause sprinklers along the line to go off one after the other, creating a very high water demand (gpm) on that single branchline, which in turn creates higher friction loss. This of course means larger pipe sizing. If the fire grows perpendicular to the branchlines opening the last sprinkler on each line then you would only have a single sprinkler flow on each branchline which would create a much smaller friction loss meaning smaller pipe sizing. Now, go ahead and ask… do we know which way a fire is going to grow? No. So, we create the most hydraulically demanding area in which we are going to simulate every sprinkler inside of that boundary (with a few exceptions) flowing at the same time from minute one.
Number of Sprinklers Activating
This leads us to our second built-in safety factor. Do all the sprinklers inside of the area of operation all go off at the same time from minute one? Answer –no. Of course, there are reported events in which all the sprinklers did go off in the area of operation and, in some cases, even more. But to date, I cannot find any documented event in which they all went off simultaneously at minute one. Which begs the question: why do we calculate them as if they do? Answer – safety factor. Even though the NFPA 13 Committee attempts to address installation issues with every edition it publishes, there will always be those situations that cannot, or will not be addressed. We calculate systems simulating simultaneous activation for this very reason. We know there are countless issues that arise in the built environment that affect either spacing, location or spray pattern. Trying to address them all would be futile. So, we activate them simultaneously to create a safety factor that, so far, has served us well. We are still reporting statistics that only one to two sprinklers go off in 88 percent of reported activations (wet systems) and 73 percent (dry systems).1 These numbers validate this design requirement and will most likely stay this way until the numbers do not support it. So, remember that as you agonize over 0.5 psi in your calculations.
ADD vs DD
The third reason our degree of accuracy should not be carrying decimals is because of what the actual pressure and flow is discharging from the sprinkler(s), especially the first few. What I am eluding to is the amount or density of water actually spraying out of these sprinklers. The terms I am referring to are design density versus actual delivered density (DD vs ADD). Remember that the calculation procedure you are performing is a “demand” calculation. This means you are creating a customized pipe schedule that optimizes your system to the most efficient pipe sizing schedule possible while staying under your supply curve. I often explain this by saying you can perform hydraulic calculations without having any water flow data. I do not need to have a supply curve to do hydraulic calculations on my system unless, of course, I want to know if it will work with a given water supply or not. It is then that I need a supply curve to plot my demand point under or over, depending on your pipe sizing. Our demand calculations are creating a minimum condition. Therefore, knowing that the sprinklers are not going off all at the same time, but rather one at a time, we can reason that the actual amount of water discharging from that first sprinkler is going to be much higher than the minimum that we created our demand calculation with. For example, if we set up a calculation that is based on a minimum density of 0.10 gpm/ft2, we are saying that if all the sprinklers inside of our area of operation go off at the same time, and the water supply is exactly as we have entered it into our calculation program, you remember, the one with the bouncing needle, we should be seeing densities of not less than 0.10 gpm/ft2 from any one of the sprinklers. However, if only one sprinkler activates within this area of operation, then there will be considerably more water at a much higher pressure than our minimum demand condition. We recently had a project that required a performance-based design solution. The minimum design criterion was 0.19 gpm/ft2 over an area of operation of 2,600 ft2. The minimum demand calculation required 25 gpm at 10 psi. We then performed a supply calculation to see what was actually going to be delivered to the sprinkler which came out to be 91 gpm at 128 psi! Using the assigned area of the sprinkler we can approximate the ADD for this first sprinkler to be 0.82 gpm/ft2 which is 332 percent over the minimum DD of 0.19 gpm/ft2. This further proves that decimal places in our calculations are not nearly as important as we perceive them to be.
So, how close is too close? 10 percent? 10 psi? We calculate our systems trying to squeeze every dollar we can out of them, hoping they will fit within a few extra 90° ells. We draw a line on graph paper and then are told to stay under it. All the while we have calculations printing out values to the hundredth of a decimal place and a water supply needle bouncing with a range of 10 psi. I will offer my opinion, but one thing is for sure. Whomever is determining what the margin of safety should be (the difference between the demand dot and supply curve) needs to have liability insurance. NFPA 13 intends for that to be the engineer of record (taken from the owner’s certificate that is required to be provided in Chapter 4) which is interpreted as the owner’s representative. In fact, something new to the 2019 edition will be the addition of this specific information added to the owner’s certificate requirements.
Regardless, if the specifications say 10 percent or the Authority Having Jurisdiction (AHJ) says 10 psi, or your company has a policy to follow, you should remember that safety margins are influenced by built-in safety factors that should be reached using sound engineering principals including acknowledgement of the actual occupancy or hazards within, not just the name of the project in the title block. I personally like the pressure approach rather than a percentage. Pressure margins are more tangible to me especially when evaluating how close it will fit. Rounding to the nearest psi has served me well for many years, especially when I know what the ADD is; what the other built-in safety factors are; and, of course, how tight I have to hold on to the pitot tube while the needle bounces back and forth!
About the Author: Steven Scandaliato is executive principal at SDG, LLC. He has over 35 years’ experience in fire protection engineering and design covering all types of fire protection and life safety systems. He serves as a principal member of the NFPA 13, 101 and 5000 committees. Scandaliato is published in several periodicals including articles for the NFPA Journal, Fire Marshals Quarterly and ASPE Journal (American Society of Plumbing Engineers). He is also a contributing author to the text published by NFPA/SFPE titled “A Designer’s Guide to Automatic Sprinkler Systems.” Over the last 18 years, Scandaliato has presented seminars to thousands in contracting and professional associations including AFSA, ASPE, the American Institute of Architects, the Society of Fire Protection Engineers, the American Society of Sanitary Engineering and the International Fire Marshals Association. He is a member of AFSA, NFPA, and SFPE.
EDITOR’S NOTE: It should be noted that the above is the author’s opinion as a member of the NFPA 13 Installation Criteria Technical Committee. It has not been processed as a formal interpretation in accordance with the NFPA Regulations Governing Committee Projects and should therefore not be considered, nor relied upon, as the official position of the NFPA or its committees.