In 1981, the second- and fourth-story suspended balconies in the Hyatt Regency Hotel in Kansas City collapsed and killed 114 people and injured 200 others. Investigation determined that the catastrophe resulted from a seemingly simple, but fatally flawed, field change that did not get the analysis it needed.

I’ve done many exercises showing the proposed change to people in technical fields and without referring to the catastrophe, asking them if they felt that the change would affect the integrity of the design. Most generally believe that the change would have no effect on the loading of the walkways. Similarly, I’ve shown that same change to freshman-level students, but direct them to take two or three minutes to draw it out, label the forces, and run the numbers before answering. Almost without exception, freshmen in an introductory-level statics class can easily see the huge effect of the proposed change. The design change was apparently accepted by experts in the field but was not accompanied by a simple analysis of the forces, and that shortcut cost 114 innocent victims their lives in one of the most catastrophic engineering failures in recent history. A simple freshman-level analysis could have averted the problem entirely.

Something happens when we leave the classroom and move into the field. We face a very different environment from where we learned the fundamental theory, whether in the university or a professional training seminar. Maybe we abandon the analysis because the real world is loud and dirty, fast-moving, and seemingly very different than the “textbook” environment we were schooled in. Maybe it’s because it’s so demanding and we don’t feel we have the time. There are no “givens” printed out in neat order with an accompanying diagram. When we begin work in a plant environment, we are immediately surrounded by people with tremendously more experience than we have ourselves; they seem to have accepted ways to work that don’t include a lot of book-work, and we begin to rely on intuition, rules of thumb, and accepted experts in the field.

There is nothing wrong with seeking opinions and guidance from those around us. It is also fine for us to use our own judgment and common sense. The problem occurs when we simultaneously abandon all education that came before, or when we stop taking the time to re-visit the basic principles, and yes, actually run the numbers.

The human mind can be very effective in making good judgments and drawing good conclusions based on intuition or common sense. Remember though, that the human mind does not process parameters that are not linear very well at all. A person estimating the total interest paid on a 30-year mortgage rarely gets close because compound interest is exponential rather than linear. The holder of a lottery ticket often has a completely unreasonable sense of hope because probability factorials are certainly not linear.

For phenomena that are non-linear, we think our intuition is much better than it really is. Are our piping systems linear? Substituting a 2-inch line for a 4-inch line doesn’t double the velocity, it quadruples it. Quadrupling velocity with a given friction factor doesn’t quadruple the energy loss, it increases it by a factor of 16. Increasing energy losses in a suction line doesn’t linearly degrade a pump, but rather it sets into place many other non-linear results that drive bearing load (a squared relationship), bearing life (cubic), and cavitation ( a suddenly crossed threshold). Fluid systems are non-linear and our judgment and intuition is not nearly as good as we think it is. We must “run the numbers” to make good decisions.  

Consider three quick samples that might arise in your facility:

1. A system with a horizontal run is designed with 1.5-inch steel pipe to carry 90 GPM of water. Due to a minor oversight, it was installed with 1-inch line. Since the line is difficult to see and the OD is 1.3 inches anyway, the change has gone unnoticed since installation. The decision is made to leave the line in place given the difficulty of pulling and reinstalling what is nearly the same size pipe anyway.

Result determined by running the numbers: The head on the pump goes from a designed value of about 50 feet to over 200 feet with the smaller pipe! At the same time, the pressure drop across the line goes from about 20 PSI to over 90 PSI. The effect on the pump is significant, but even if a different pump is installed to match this system, power consumption increases by 2.7 kW, more than a factor of four, the entire time this system is operated.

2. A pump is installed next to a column, and a look at the original drawing indicates that the column was not accounted for during the design process. What was to be a short-straight suction line from the tank became a series of four elbows going around the column, but the system seems to work now and there really is no easy way around stationary columns.

Result determined by running the numbers: The designed suction line had less than a foot of head loss, and it was replaced with one that introduces losses of over three feet! This difference takes the application from a robust design to one that lacks adequate net positive suction head available over some of its operating range, resulting in cavitation, in addition to any effects that result from an inadequate straight run to the pump.

3. An application runs in two different modes, one at 600 GPM and one at 300 GPM. You notice that a pullout pump is in place, but the impeller is never changed. Your research indicates that initially changes were made between seasons, swapping the 10-inch impeller for a 7-inch.

Maintenance and production met at some point, however, and determined that the time to make the change, both in labor and lost time, are simply not worth it. The larger impeller works for both settings. The change is rather minor anyway and the smaller flow can be achieved with an easy valve adjustment.

Run the Numbers: Do's & Don'ts

DO: Examine the original drawings and compare those to the actual physical systems.
DO: Review the product literature for the commercial components in the system. Find K values, Cl values, equivalent lengths, etc.   
DO: Generate system curves—the pump curve is meaningless without the system curve.
DO: Install gages, at least at pump inlets and outlets to allow analysis and active operation of pumps.
DO: Get buy-in, gather data, make calculations, bring in others again to review the actuals, take action to correct and improve the system.
DO: Create a culture that embraces real numbers and diagrams—not as nerdy, theoretical, and irrelevant, but as the key to improvement.  
DO: Consider software that is capable of making calculations and modeling your system

DON'T: Rely on any single source of information: sales rep, engineer, operator, skilled trade.
DON'T: Ignore any source of information (same as list above)!
DON'T: Let a “rule of thumb” contaminate the work (“I heard that these meters always have a 5 PSI drop” or “the system runs at half flow at night” etc., etc.
DON'T: Think that analysis is only for the design of new systems. Just talk to people and then apply your our own general intuition. Seat of the pants and intuition are not serving industry well.
DON'T: Wait for absolute perfection—you don’t need to gauge to .0002 percent before you can analyze and take action!

Result determined by running the numbers: When the lower flow condition was achieved with an impeller change, the system operated at BEP and 30 feet of head, consuming just under 5 horsepower. Substituting the impeller change for the simpler method of adjusting a throttling valve, the system was able to give the required conditions and operates at nearly the same efficiency, but operated significantly away from BEP, degrading the life of the pump dramatically and consuming an additional 15 HP that is all wasted.

For the situations above, a failure to do the simple analysis means the cause and effect, made so evident with basic analysis, would not be revealed. In lieu of analysis then, our reaction might be to tolerate poor performance, waste energy resources, add double-seals, or monitor vibration more closely. Without the analysis, we would forego the opportunity for an irreversible corrective action that only good analysis and follow-up action can bring.

There are multiple acceptable approaches to completing the analysis of a system. Alternatives include simplified techniques you might pick up in a training program or a manual, theoretical material from college texts, and there is also software available designed specifically to make these types of analyses simple and intuitive. Any of these approaches is superior to accepting relative ignorance.

Often times when we neglect the quantitative analysis there are not catastrophic results. In other words, most times when we replace engineering analysis with intuition, conventional wisdom, or unintentional neglect, we don’t kill 114 people. But many times since 1981, your facility probably could have fared better if someone took the time to draw up your actual systems and run the numbers. In Kansas City about 30 years ago, if someone would have taken a few minutes to do a freshman-level analysis, 114 lives could have been spared. How far off are your implicit assumptions about your own piping systems? Run the numbers, and you’ll find out.

Joe Untener is a professor of Mechanical Engineering Technology at the University of Dayton. A registered professional engineer, Joe has 25 years of engineering experience in industrial and educational settings, having worked not only for General Motors, but  also as a consultant in dozens of industrial plants throughout the United States. He is co-author of the 7th Edition of Applied Fluid Mechanics with Robert L. Mott, due out through Prentice-Hall in 2013. Mr. Untener can be reached at