|| Larry Cottrill |
| Date :
|| 2002-01-23 12:25:39
| Subject :
|| Mini Myers Initial Tests -- Long Post
Bruno, Graham et al -
Yesterday afternoon, I was able to test fire my tiny Myers (or quasi-Myers) model. Basically, it is a carbureted modification of the Myers design, using a 'flowjector' type air/fuel delivery system, supplying forced air and fuel spray for starting -- but, traditionally, forced air is the only way my little engines ever run, and this example turned out to be no exception! However, a few things have been observed which may be of interest.
I have previously stated that I believe that the 'choking' effect is easily achieved, without resorting to sophisticated design of the nozzle, if the impinging gas velocity is high enough. I decided that the only thing I really wanted to test for the moment was whether I could use Myers' arrangement of a front-end 'nozzle' (what I have called the 'Myers choke') and tail-end ignition to achieve pulsation, thereby validating the choking effect as a reflective 'front plate' for the explosions. Because I have no fuel injection equipment (and am really not personally very interested in injected engines, anyway) a means of carburetion is substituted for the injector and mixing chamber Myers shows (I believe the mixing chamber could still be used, aft of the carburetor, and that's the way I originally intended to build it, but decided to do without it as a 'first shot' at the problem). Because of the intuitive judgment that smooth flow from the carburetor nozzle would be necessary for good air delivery and because the mixing chamber is abandoned, I also got rid of Myers' 'sharp edge' at the beginning of the nozzle zone. If Bruno's theory of the sharp edge effect is correct, it would only be important in the case of a low-velocity 'mixing zone' anyway, in my opinion. So, the 'Myers choke' nozzle simply becomes the exit skirt of a carbureting intake venturi. Only one spark plug is used, since the tube's internal cross-section is quite small.
These departures from slavishly duplicating Myers' front-end design will certainly be criticized, but I defend it on this basis: The only thing I wanted to learn is whether full choking action against forward flow from the ignition zone is achievable, and especially whether it is achievable without resorting to refined design theory. Anything learned beyond that will be 'icing on the cake'. Hence, exact duplication of the Myers plan is not considered a requirement, for now.
I started with a piece of 1/2 inch EMT (Electrical Metal Tubing or 'rigid conduit') 15-1/2 inches in length, with both ends filed smooth and free of burr at the cut edge. (The actual ID is just over 5/8 inch.) 3/4 inch of the front end was placed in the bench vise, which was gradually tightened until deformation resulted in an oval approx. 1/4 inch across, inside. So, at this point, a flat 'duckbill' nozzle had been formed. Next, two 3/8 inch twist drills were secured opposite each other across the flat sides of this 'duckbill' using rubber bands, such that the smooth shanks of the drills were held parallel to one another and as far rearward as they could rest on the flat faces just formed. The vise was opened up to accommodate the entire width of the drill-and-tube 'sandwich' and tightened while making sure that the drills stayed parallel to one another and perfectly transverse to the tube centerline. The vise was gradually closed until a 5/32 inch drill shank inserted in the gap was just seized, then the vise was relaxed. So we now had a crude venturi with a 5/32 inch wide slot (about 3/4 inch high) with smoothly rounded contours for the throat, expanding forward to an approx. 11/16 x 1/4 inch oval intake and rearward to a 5/8 inch ID tube, which is our 'combustion tube' section. A point 4 inches from the other (rear) end was marked, and a steel nut fully welded in place at that station and at a point which would be the designated underside of the finished tube. (This was done with the oval intake port centerline assumed 'horizontal', like the intake of the old F-100 jet fighter). A mount plate to fit my little steel test jig was fabricated from 1/16 inch sheet steel and welded on, carefully aligned just aft of the deformed nozzle zone. The welded nut was then drilled out to penetrate the underlying tube wall and threaded with the 1/4-32 NXF tap, to accommodate the Champion H-3 (or V-3) spark plug used. Oval slots were filed into the intake edge with a round needle file to fit the air/fuel pipe assembly (the same 'flowjector' I used in the Synchrodyne tests) and carefully adjusted so the fuel ports are almost perfectly centered between the facing rounded edges of the venturi throat. Note that it was impossible to center this in the 3/4 inch dimension, but almost perfect centering within the narrow width of the throat was easily achieved by fitting, inspecting and re-filing. Final mounting of the air/fuel pipe and mounting the plug made the engine complete. I mounted the finished engine on the test jig and provided a new piece of clear vinyl tubing between the 1-1/2 ounce fuel tank and the fuel inlet -- the device was ready to test.
TEST METHODOLOGY -
My starting rig consists of a Model T Ford spark coil with intermittent push-button switch and a small air pressure tank with adjustable pressure regulator and intermittent push-button valve. I decided to start with a low or moderate pressure, and work up as needed while watching for pulse operation and observing the intake for indications of forward flow. Adjustments of the needle valve in the air/fuel pipe assembly would be made as needed for good combustion.
Since I haven't had time to sit down and review my test videos, I won't try to detail the test variations, moment by moment, but rather just summarize what I can remember as basic observations:
- With appropriate needle valve settings, it is possible to obtain good smooth combustion at almost any air delivery pressure and flow, if you're willing to settle for continuous (non-pulse) combustion.
- Unlike previous experience with my other designs, there seems to be no tendency for the flame front to separate from the end of the tailpipe under very high-flow conditions, even with a fairly rich needle valve setting. This seems to hold true whether spark ignition is maintained or not. So far, I have not thought of a satisfactory explanation for this difference.
- Unlike the Synchrodyne tests, in this case it seems to be impossible to obtain pulsing combustion at low rates of flow. Although the volume of this tube is just a small fraction of the volume of the DynaJet, a similar airflow is needed to achieve explosive combustion, with a correspondingly high fuel consumption!
- It appears necessary to use a fairly fuel-rich mixture to achieve fully pulsing combustion, although a suggestion of pulsing is audible at leaner settings.
- Regardless of variations in air flow or other test conditions, there was never any flame observed flowing forward through the throat of the nozzle, nor was there any other indication of reverse flow at any time. This is in sharp contrast to the Synchrodyne tests, where significant forward flow during pulsing operation was easily observed.
- There was virtually NO 'popping' or 'banging' observed -- if the mixture was rich enough to burn, either smooth combustion or pulsing began immediately when air and spark were applied. This is quite different from test observations in my other designs, and from common experience with starting the DynaJet.
- There is a fuel delivery problem at high flow settings: If there is sufficient fuel and air to achieve pulsing, a small amount of liquid fuel accumulates in the combustion tube and trickles rearward and drips from the tailpipe end, for a dramatic and colorful 'special effect'. I have a hard time believing this is from poor atomization, because in the absence of ignition, air/fuel spray from the tail end seems to be a perfectly uniform cloud of fine mist. I checked the centering of the fuel pipe in the nozzle after the tests, and it was still as symmetrically positioned as I can get it. So far, I haven't come up with an explanation for this behavior.
- There is no evidence of combustion in the forward part of the combustion tube, other than the fact that pulsing operation was finally achieved. Slight discoloration of the tube is visible at the very tail end, but there is no evidence of significant heat upstream of the spark plug. No single run was of sufficient length to achieve observable red heat anywhere in the tube (note that these were not nighttime tests -- lighting was filtered north sky light). The few times I thought to observe it, the engine front end was fairly cool after each test run.
- Pulsing operation, under high air and fuel flow conditions, seems like fairly impressive pulsejet output. It has the 'sharp' character of valved jet operation, not the 'dull' sound usually described for valveless designs (again, we need to remember that this is under a forced air running condition, not self-aspiration!). It sounds like the DynaJet, but of course not nearly as loud (I could hear myself yell over it, for example).
- The frequency during pulsing operation is definitely the 'closed pipe' frequency, not the higher 'open pipe' resonance frequency. This relates to my experience with the Synchrodyne tests, where both frequencies were seen to occur, with the lower 'closed pipe' operation happening when forward flow from the chamber was occurring -- but again, in this case, there is NO observation of forward flow of any kind.
- To me, the most interesting observation of all: Pulsing operation was not maintained after spark ignition cutoff -- even when good pulsing was obtained, the device immediately fell into continuous 'forced air burner' running when the spark was interrupted. The instant spark was restored, pulsing operation was resumed. Note that this is basically a 'cold tube' observation, which might change if a run was long enough to really heat the rear of the tube; however, as it stands, the effect is quite striking -- you can simply 'switch' pulsing on and off instantly with the push-button!
TENTATIVE CONCLUSIONS -
It seems to me that there is reasonable evidence that the choke is working under pulsing conditions. I can't really imagine explosive pulsing operation that doesn't involve significant forward flow in the tube, and the frequency indicates 'closed tube resonance' operation -- also, the sharp character of the sound seems indicative of reflection off a 'hard' stop, not the 'softened reflector' model of a typical 'leaky' valveless front end. Of course, there could be a lot of wishful thinking in there, too! However, (at the moment) I think the principle is satisfied, and the basic Myers idea (as clarified by Bruno) well vindicated.
The current design seems to have a basic gross inefficiency, requiring far too much fuel and air drive. It seems inconceivable that the high level of energy presently required could ever be sustained by self-aspiration in the present device. It is entirely possible that a more refined nozzle design could, in fact, alleviate much of this problem, in my opinion -- obviously, the exact nozzle characteristics are going to determine how much gas flow energy is needed to achieve choking. It is possible that my throat area is actually smaller than it needs to be, with correspondingly 'too fast' cross-section reduction from the tube toward the throat, as was alluded to by Graham in the earlier 'Myers choke' thread I started.
The ignition spark is obviously creating a pressure peak (i.e. antinode) which strongly influences what happens in the tube. Without the spark, ordinary Kadenacy re-aspiration and firing was not achieved. This engine is rather short for a very small pulsejet, so it is entirely possible that ordinary pulsejet operation would have occurred with more tube length. What it does certainly prove is that, unlike in most pulsejet designs, the presence of spark within an already established combustion stream makes a profound difference in internal behavior in this device.
Based on very preliminary test observations made on an admittedly crude model, I feel that there is reasonable evidence that a working engine can be derived from Myers' design. I still believe that optimization for a carbureted design is possible; however, someone with liquid fuel injection or propane fueling capabilities should try a small model that is somewhat larger and which more strongly resembles the original Myers pattern.
Comments from everyone are welcome, of course.