I spent several days designing a gear pump and then, just before starting its machining, I began having second thoughts. The Offy's coolant system is a pump located in the bottom of a closed loop that's completely filled with coolant. Since the water columns at the pump's inlet and outlet are at the same height, it should be capable of circulating the coolant with minimum pressure across it.
Wall friction within the tiny head passages in the upper part of the engine could conceivably create a flow bottleneck and increase the pressure requirement of the pump, but this can be calculated using Poiseuille's law:
Q = (pi * r^4 * P) / (8 * n * L) ,
Where Q is the flow rate,
P is the pressure drop,
r is the radius of the restriction,
L is the length of the restriction, and
n is the fluid's viscosity.
Plugging in values for an 1/8" diameter restriction that's 6" long (typical of one of the Offy's head passages) and using the viscosity of water shows that only .0023 psi is required for each cubic inch per minute flow. Full-size automobile engines typically turn their entire coolant volume over one to two times per minute. For the same turnover rate, the Offy's 3 to 6 cubic inches per minute should require no more than .014 psi from the pump. (For a sanity check on the math, visualize blowing through a short soda straw.)
It doesn't seem likely that removing waste heat from the Offy's head can be significantly improved with a gear pump capable of producing tens of psi pressure. It's more likely that the pump would have to be severely throttled back to prevent coolant leaks from an already questionable head gasket.
Setting aside my first attempt, a well-designed centrifugal pump should have little problem circulating coolant through the Offy. I carefully considered the comments received on my first pump before taking another stab at it. I improved the tangential exit and water cut and added some semblance of a real volute. Since there was a definite advantage to retaining the same impeller diameter, space for the volute was created by increasing the diameter of the pump body and then notching it for clearance around the starter shaft. Although still not ideal, the volute's geometry was considerably improved. The height of the impeller was also increased by 30%. I don't normally like making so many simultaneous changes while working my way up a learning curve, but it's getting time to move past this part of the project.
A few material changes were also made to improve the pump's long term corrosion resistance. It turns out that aluminum and stainless weren't the best metals to put into wet contact. Both the pump body and cover are now 7075 aluminum with the impeller was machined from Delrin. An integral Delrin sleeve also replaced the front ball bearing, and the impeller can now limit its own thrust with minimum wear to the cover. The coolant will eventually become a 50/50 mix of anti-freeze and water which will provide some lubrication and corrosion resistance.
The number of impeller blades was reduced from seven to six and their thickness increased some 30%. The impeller has a 1/4" diameter pressed-in metal shaft with a rear end that remains supported in a ball bearing similar to the original pump. Silicone grease packing and an o-ring in front of this bearing makes up,the pump's rear seal.
Since I previously saw a performance improvement with fish-mouthed impeller blades, I machined two impellers for this pump. One has full height blades, and the other one is fish-mouthed. Research showed the backward curved blades I'm using (or else offset straight blades) are the best performers for non-compressible fluids. Without CNC capability the extra machining effort required for curved blades probably isn't worthwhile.
Testing...
The new pump performed much better compared with my first attempt. It visibly circulated water through the simulated loop at effective crankshaft speeds as low as 500 rpm. In this pump, the full height impeller performed 30% better than the fish-mouthed impeller. At an effective 5000 crankshaft rpm, the pump was capable of producing a 12" water column (0.43 psi) with the full-height impeller and a 9" column (0.32 psi) with the fish-mouthed impeller. While producing a 3" water column, the full-height impeller flowed 54 cubic inches/min at an equivalent 5000 crankshaft rpm. At 500 crankshaft rpm it produced .040" psi while pumping water at 5 cubic inches/min. Even at such an optimistically low idle speed, this new pump should have no problem pumping coolant through the Offy. - Terry
