Peter Nachtwey
Member
Last year at about this time I got a call from Sandia Labs. They had problems controlling a diesel fuel injection system. I was contracted to do a simulation to figure out what was wrong with their system and provide information that would help them setup their system in the future. I did a simulation for the diesel fuel injection system. It had to compress fuel from about 150 psi to as high as 43,000 psi. The high pressures were a new one for me. There were two diaphragm pumps that are use hydraulic cylinders to pump fuel using a 1/8 piston. This allows intensification of the pressure. I used cascaded controls. The outer loop controlled the pressure and the inner loop controlled the rate at which the two cylinders make the pumps cycle back and forth.
I found a lot of problems with the hydraulic design. I could have designed the system to reduce the pressure ripple to almost nothing but they already had the equipment. I determined they had to scale back the volume of diesel fuel they could expect to pump at 43,000 psi and then went on to make it work with what they got. It took about 14-15 hours to do the simulation but it worked well. The value I predicted were close enough where they had no trouble getting the system to work.
One of the things that was new is that the fuel would compress by 30% while it was pressurized to the injection pressure. This means the pump didn't pump anything at all for the first 30% of the stroke until the fuel pressure allow the fuel to go through the check valve. I think the original designer didn't take this into account. It is obvious to me the hydraulic designer didn't simulate the system.
http://machinedesign.com/hydraulics/electrohydraulics-drives-engine-research
The purpose of posting this is that one CAN simulate systems and predict performance values and PID parameters pretty closely even if the application looks pretty complicated and is non-linear.
I found a lot of problems with the hydraulic design. I could have designed the system to reduce the pressure ripple to almost nothing but they already had the equipment. I determined they had to scale back the volume of diesel fuel they could expect to pump at 43,000 psi and then went on to make it work with what they got. It took about 14-15 hours to do the simulation but it worked well. The value I predicted were close enough where they had no trouble getting the system to work.
One of the things that was new is that the fuel would compress by 30% while it was pressurized to the injection pressure. This means the pump didn't pump anything at all for the first 30% of the stroke until the fuel pressure allow the fuel to go through the check valve. I think the original designer didn't take this into account. It is obvious to me the hydraulic designer didn't simulate the system.
http://machinedesign.com/hydraulics/electrohydraulics-drives-engine-research
The purpose of posting this is that one CAN simulate systems and predict performance values and PID parameters pretty closely even if the application looks pretty complicated and is non-linear.