By Peter Ommundsen
A popular hobby is that of building steam-driven whistles that can be sounded at steam fairs, traction engine rallies, steam pageants, and similar events. Steam whistles were once common on factories, steam locomotives, and steamships.
A common source of advice on whistle-building is the Internet, but unfortunately it is littered with whistle design claims that can mislead people into wasting considerable time and money. Below are several suggestions for navigating these claims. Items in quotation marks were culled from actual Internet postings.
ARE YOU PROMISED THE IMPOSSIBLE?
Whistle-building can be costly and time-consuming, therefore whistle design plans should be founded on well-researched principles and properly tested prototypes. Web sites, including forums and discussion groups, may offer formulas or spreadsheets giving steam whistle plans purported to yield particular pitches or musical chords when blown on steam or compressed air at specified pressures. These plans should be scrutinized carefully.
Promises for the acoustic quality of a novel whistle design should allow for latitude in whistle performance, as whistles are fickle instruments and it is not uncommon for two seemingly identical whistles blowing side-by-side to sound somewhat different. Also, blowing conditions such as temperature and steam quality can vary among venues.
Claims of perfection, such as “I guarantee that by using these spreadsheets you will turn out the perfect whistle the very first time” should be seen as red flags, for as we shall see, there have been some spectacular design failures.
DO FREQUENCY PREDICTIONS APPEAR TOO GOOD TO BE TRUE?
As any musician or calliope-tuner knows, the speed of sound is temperature-dependent, steam varies in dryness, and pitch cannot be guaranteed. And the pitch of whistles blown on steam cannot be predicted from their pitch on air or vice versa through any simple formula. A recent inventory of 14 whistles blown on both air and steam showed frequency differences between steam and air that varied from five percent to over forty percent (Horn and Whistle Magazine No. 126). Whether or not a frequency prediction is lucky may depend upon prevailing conditions. Despite this, we still see claims such as “frequency prediction will be within 1% for air and within 5% for steam.”
Several years ago a member of the Yahoo Steam Whistles Group employed a spreadsheet to design a large railway locomotive whistle. He was told that “the whistle will work exactly as planned at your intended operating pressure of 200 PSI steam. It will be very loud, in tune and will not overblow.” In fact the whistle sounded fundamental frequencies an average of 25 percent higher than forecast and the musical intervals between the notes were not as predicted. This whistle has recently been touted on the same forum as “one of the most significant breakthroughs in whistle design in the past 150 years.”
This frequency formula is based upon whistle length, but the effective length of a whistle is shortened as blowing pressure rises, and this is especially so in squat whistles. Therefore the formula should be varied with whistle scale, which has not been done. For this reason alone, frequency prediction accuracy can vary by fifteen percent or more for two whistles of different scale fully blown side-by-side.
CAN CLAIMS OF VERY LOW PRESSURE OPERATION BE DOCUMENTED?
There are countless internet postings claiming that large conventional whistles will fully blow at low pressures such as 15 PSIG if only the steam slot in the floor of the whistle mouth is large enough, e.g., 0.0625-inch for a 3:1 length/width scale whistle.
In fact many large antique ship whistles, which typically used pressures of 100 to 300 PSIG, had 0.0625-inch or larger steam slots. If the low pressure hypothesis were accurate, think how much steam could have been saved if the steam supply had been down-regulated to a mere 15 PSIG! If only those engineers had been smarter.
Hobbyists have tried without success to adjust large conventional whistles to fully blow at 15 PSIG. One enthusiast built three huge examples, 7, 8, and 10-inches in diameter of the 90-degree-mouth organ whistle design using the 0.0625-inch steam slot. These whistles fully blew at 150 PSIG or higher, not the claimed value of 15 PSIG.
Until the proponents of the low pressure hypothesis can demonstrate 15 PSIG full acoustic output of a series of large conventional whistles, these claims should be withdrawn.
Another commonly encountered claim is that the whistle mouth area and cross-sectional area of the sounding chamber should be equal (unity). For example, the mouth height of a plain single-note 360-degree mouth whistle should be one quarter the bell diameter. This contradicts early research showing that for a fixed blowing pressure the optimum mouth area for maximum sound level varies with the whistle scale.
The unity mouth has also been questioned in recent decades by hobbyists using sound level meters. Sound output may increase when the mouth area is raised above unity, perhaps at some point providing the best match of driving frequency and resonator frequency. Also adjusting the cut-up can greatly improve whistle sound quality, which can be harsh and off-key using the "unity" cut-up with its supposedly ideal slot width and pressure. Whistle builders would do well to examine traditional antique whistles, which often used a cut-up greater than unity.
ARE WHISTLE DESIGN SPREADSHEETS LARGELY SPECULATION?
Whistle spreadsheets attempt to match steam slot width (steam flow to the whistle) to a pressure that will allow a whistle to be fully blown. (Meaning that blowing pressure can be reduced without loss of sound level if the slot width is increased by a specified amount.) In the case of the Yahoo Steam Whistles Group this is done with a simple formula: a quartering of gauge pressure requires a doubling of slot width.
The formula is questionable for three reasons:
(1) The claim is that reducing pressure but widening the slot maintains the same gas flow to the whistle. But flow to a whistle in the normal working pressure range is proportional to absolute pressure, not proportional to the square root of the gauge pressure as used in the spreadsheet.
(2) The pressure required to fully blow a whistle for a given slot width is not constant for all whistles of the same scale. For example, pressure requirements vary with mouth type (organ or plain), size (tiny or huge), and whether blown on steam or air, which have different viscosities and different densities. One formula does not fit all.
(3) Spreadsheet predictions of particular pressure/slot-width trading lack agreement with the results of field tests.
A good whistle spreadsheet would benchmark the formula with an abundance of comparative sound-level readings from field-tested whistles blown side-by-side. These data should confirm that reducing the pressure and widening the slot by the prescribed amounts preserves the sound level of the whistles.
If abundant field data from a range of whistle sizes and scales are not provided in the spreadsheet, users should be wary, should assume that the spreadsheet is speculation, and try to test the formula with an existing whistle. This can be done by testing languid plates of various sizes to check the supposed decibel equivalence in trading between specified slot widths and blowing pressures.
JOE LIKES HIS WHISTLE
A common retort to all of the above that evades addressing the issues raised, is that a particular spreadsheet is fine because someone made a whistle that “works” and he likes it.
A “working” whistle does not mean that "the spreadsheets work," because that whistle is not a test of the many other whistle design combinations that may be embedded in the spreadsheet, many of which may not work as claimed. A "working whistle" does not validate questionable claims of the spreadsheet as a whole, including slot width / pressure trade-offs that would supposedly enable fully blowing a huge whistle at 15 PSIG, accuracy of frequency predictions over a range of whistle scales, and the efficiency of a unity mouth.
ARE QUESTIONS AND NEW IDEAS WELCOME?
Before joining an Internet whistle forum, determine if questions are welcome, if debate is fosterd, and if the physics behind the formulas is transparent. In some cases the forum may be simply a platform for promoting tenuous theories that members don’t want scrutinized and collective wisdom is unwelcome.
For example, I noticed recently that a member of the Yahoo Steam Whistles Group who questioned their whistle design math was evicted rather than encouraged to share his knowledge and concerns. Such censorship makes a mockery of the supposed educational mission of the group and has a chilling effect upon knowledgeable members who might otherwise help improve whistle plans for the benefit of consumers.
Fortunately there are other forums that do welcome transparency and debate, such as the Yahoo Horn/Whistle Group and the Horn and Whistle Board.
MUCH TO LEARN
There is much that is unknown about steam whistle physics and there is a certain risk in betting on a novel design. The beginning whistle builder might do well to copy a tested design. Purveyors of whistle design formulas would do well to raise awareness about the uncertainty of untested designs, to provide ample field test data from prototypes, and to update formulas when shown to be flawed. References to some actual field tests can be found at the Wikipedia steam whistle page.
The following quotation from a 2006 article in the engineering journal Experiments in Fluids equally applies to whistles. The emphasis is mine.
November 8, 2011