A few small parts later, and I had a functional water pump. The impeller was machined from Delrin and pressed onto a 3/16" shaft. Its one thou interference fit wasn't tight enough to keep it from slipping under load and so it had to be pinned. The shaft runs in a pair of 3/8" sealed ball bearings separated by a half inch spacer. It's sealed by a pair of silicone o-rings located on the shaft below the rear bearing. For good measure and lubrication while dry running, the o-rings were coated with silicone grease. Drag on the spinning impeller was kept to a minimum by insuring its two thrust surfaces were free of grease. The pulley mounting flange was secured to the shaft with a grub screw tightened against a machined flat. Axial thrust is limited by the impeller itself spinning between the backing plate and an integral thrust bearing machined into the pump body.
At least two varieties of impellers could be found on the small block Ford. A six vane stamped steel impeller was the factory's cost effective solution. A 'high performance' cast aluminum version was also available from the aftermarket. One difference between the two is that the stock impeller has straight vanes while the aftermarket version has curved vanes. Out of curiosity I decided to machine versions of each and compare their relative performances in a crude test bench. For some reason, maybe because it was fun, I made spares of both impellers.
The coolant outlet tubes double as dowels to locate the water pump and timing cover to the block. They fit snugly inside the pump's backing plate, but their orientation inside the pump body is very important. Both ends of the tubes were identically beveled so their orientations in the pump sub-assembly can be verified. The dowels are sealed by the pump and cover gaskets.
The outlet tubes were fashioned from quarter inch aluminum hobby tubing, but with a wall thickness of only .014" they are rather fragile. I was concerned that over time they might become stuck inside their close-fitting surroundings and create problems for later disassembly. Anodizing or nickel plating would reduce the long term effects of their wet environment, but based upon previous experience I had concerns about making reliable electrical connections to them during plating, Instead, I settled on an alodine dip but just in case I also made an extra pair.
The pump and its impellers were tested as those were in my Offy build. The pump's input was fed from a water reservoir approximating the expected volume and input/output hose locations of a radiator. The pump's output hoses dump into a 1500 ml beaker. With a drill spinning the pump, the time required to empty the reservoir into the beaker was measured. In a second test, the output hoses were raised above the water level in the reservoir until the flow stopped giving the maximum water column supported by the pump. This number was converted to a rough head pressure measurement.
My best Offy results were used as a baseline for comparison since that pump worked well in actual operation with the engine's 3 cubic inch coolant capacity. At 1300 rpm it flowed an average 15 SCIM (std. cubic inches per minute) and was capable of .32 psi head pressure.
The stock six vane impeller running in the HI-PO's pump body was tested first. It produced the same .32 psi head pressure, but flowed a whopping 115 average SCIM. I expected some improvement over the Offy's pump because of the V-8's larger scale and dual outputs, but the 7.5X improvement was a pleasant surprise. The test was repeated several times with consistent results.
The curved vane disk impeller was tested next. It was the clear winner producing .43 psi and flowing an average 170 SCIM. Since the Hi-Po's estimated coolant capacity is 8.5 cubic inches, this pump should work well assuming the actual flow rate achievable in the engine is any where near the pump's capability.
Another positive result was that after the few days of testing there were no noticeable leaks. I did discover though that I'd made the inlet barb from steel instead of stainless when the water in the beaker began developing a brownish tinge.
I had planned to start on the crankshaft next, but will instead begin work on the bell housing. It would be nice to finish up the major parts requiring marathon machining sessions before the summer heat sets in, but to be honest I don't like making crankshafts and I'm probably procrastinating. In any event, I haven't yet created a model for the bell housing, and so it looks like I'm headed back to SolidWorks for a while. - Terry
At least two varieties of impellers could be found on the small block Ford. A six vane stamped steel impeller was the factory's cost effective solution. A 'high performance' cast aluminum version was also available from the aftermarket. One difference between the two is that the stock impeller has straight vanes while the aftermarket version has curved vanes. Out of curiosity I decided to machine versions of each and compare their relative performances in a crude test bench. For some reason, maybe because it was fun, I made spares of both impellers.
The coolant outlet tubes double as dowels to locate the water pump and timing cover to the block. They fit snugly inside the pump's backing plate, but their orientation inside the pump body is very important. Both ends of the tubes were identically beveled so their orientations in the pump sub-assembly can be verified. The dowels are sealed by the pump and cover gaskets.
The outlet tubes were fashioned from quarter inch aluminum hobby tubing, but with a wall thickness of only .014" they are rather fragile. I was concerned that over time they might become stuck inside their close-fitting surroundings and create problems for later disassembly. Anodizing or nickel plating would reduce the long term effects of their wet environment, but based upon previous experience I had concerns about making reliable electrical connections to them during plating, Instead, I settled on an alodine dip but just in case I also made an extra pair.
The pump and its impellers were tested as those were in my Offy build. The pump's input was fed from a water reservoir approximating the expected volume and input/output hose locations of a radiator. The pump's output hoses dump into a 1500 ml beaker. With a drill spinning the pump, the time required to empty the reservoir into the beaker was measured. In a second test, the output hoses were raised above the water level in the reservoir until the flow stopped giving the maximum water column supported by the pump. This number was converted to a rough head pressure measurement.
My best Offy results were used as a baseline for comparison since that pump worked well in actual operation with the engine's 3 cubic inch coolant capacity. At 1300 rpm it flowed an average 15 SCIM (std. cubic inches per minute) and was capable of .32 psi head pressure.
The stock six vane impeller running in the HI-PO's pump body was tested first. It produced the same .32 psi head pressure, but flowed a whopping 115 average SCIM. I expected some improvement over the Offy's pump because of the V-8's larger scale and dual outputs, but the 7.5X improvement was a pleasant surprise. The test was repeated several times with consistent results.
The curved vane disk impeller was tested next. It was the clear winner producing .43 psi and flowing an average 170 SCIM. Since the Hi-Po's estimated coolant capacity is 8.5 cubic inches, this pump should work well assuming the actual flow rate achievable in the engine is any where near the pump's capability.
Another positive result was that after the few days of testing there were no noticeable leaks. I did discover though that I'd made the inlet barb from steel instead of stainless when the water in the beaker began developing a brownish tinge.
I had planned to start on the crankshaft next, but will instead begin work on the bell housing. It would be nice to finish up the major parts requiring marathon machining sessions before the summer heat sets in, but to be honest I don't like making crankshafts and I'm probably procrastinating. In any event, I haven't yet created a model for the bell housing, and so it looks like I'm headed back to SolidWorks for a while. - Terry
