Advanced Engine Cooling System Control, a Reflection

My Experience

I joined Clemson’s Formula SAE team my freshman and sophomore years and was a member of the powertrain division. I investigated the cooling system of the vehicle, looking at how we could improve the heat rejection per unit area, in order to reduce the footprint and weight that the radiator and cooling system components contributed to the total weight of the vehicle. I participated in undergraduate research with Dr. Han that was centered around subsonic wind tunnels (See figure 1). I created a test stand for the radiator and measured the dynamic and static pressures in front of and behind the radiator. During these tests I learned a lot about fluid mechanics and how to use the measurement equipment with the data acquisition boards. However, I had not taken a course in heat transfer and was trying to learn on my own how heat transfer occurred at the cylinder walls all the way back to the radiator fins. I was overwhelmed at the time by the complexity of the mathematical models that were developed to describe these phenomena. I eventually was able to size the radiator and suggested implementing an electric water pump. This paper was interesting for me to read and reflect on my experiences back when I knew very little about engineering.

Figure 1. Closed loop wind tunnel at the FSAE shop.

Purpose and Expectations

My initial expectations before reading the paper [1] was that there would be a proposal of a new advanced radiator or feature that improved the radiators ability to reject heat to the air. This was based on my experience and was not the true purpose of this report. I read the abstract and immediately had a clear idea of what this report was about. Without reading any of the details of the report, I thought about how the experimental setup might be arranged in order to validate the prescribed NAMIMO controller. From my experience in a wind tunnel, I imagined a large supporting structure that would be used to mount the radiator, with the fans and fan shrouds attached. An engine would supply the heat to the cooling system and the actuators would vary the valve position as well as the fan speeds in order to regulate the engine coolant temperature. I was interested to read more about the design of the controller and understand the performance of the NAMIMO compared to the NAB and SF controllers.

Reflections

I appreciated the context that was provided in the introduction with respect to the relevance of this research and explaining how classical systems regulated the coolant flow through the radiator. It was necessary to also explain in the beginning that the aim of the cooling system is not to over cool the system, but to maintain a desired coolant temperature to the engine, otherwise performance degradation would occur. This helps readers better understand the aim of the proposed controllers and without knowing the details of the mathematics, the reader can better grasp the concepts on a systems performance level. Later in the paper, the error between the desired temperature and the measured temperature draws back to this initial piece of information and ties it altogether nicely. Figure 1 is a great 3D model that illustrates the components and is paired well with the coolant flow directions and variables provided in figure 2. Several contributions to this experimental configuration of the cooling system were mentioned but there was one that I found especially interesting. The optimal fan matrix control strategy to minimize the radiator fan power while achieving the requested heat rejection. This made me consider how I might approach such a problem, and how using different tube arrangements might work together with the fan matrix. In this case the tube would be arranged in such a way that the fans would need to operate for shorter amount of time and consume less power. In the modeling section I thought it was interesting how the equation at the junction was developed. In my mind this kind of mixing valve between the cold and hot fluids would lead to a more complex set of equations, but this captures this valve position well, and assumes that there is a proportion of hot or cold fluid equal to the valve position. Moving on I continued to read more about the control strategy itself. The controllers aim is to regulate the engine and radiator temperatures, but it is subjected to disturbances that simulate the real world. They propose a closed loop feedback system with a proportional controller with the objective being to minimize the error to zero. Without any additional temporal specifications for the controller, its not known how quickly the system would stabilize or what kind of overshoot is acceptable. In section C of control strategies, the 3rd remark highlights the importance of why the fan on time should be reduced. Not only is thermal efficiency important and the desired goal of the system, but now further consideration of the electrical power consumption should be valued in today’s ever increasingly electric world. The experimental test bench used low pressure steam in an 8-cylinder engine to simulate the heat generated by the combustion process. I thought this was interesting and unexpected. Its reasonable to use steam rather than continuously run an engine, especially in the high bay of Flour Daniel. I was really confused about how the tests were performed, it was difficult to interpret Table 1 but I continued reading and found figures 5 and 6 to be more useful. I was impressed by how well the NAMIMO controller was able to reduce the error and minimize the power consumption. The system was able to respond well to disturbances in heat load as well as changes in ram air speed. Overall, I thought that the paper provided more insight into the practical application of controllers and illustrated how there is still room for improvement in improving the efficiency of time-tested systems.

Next Steps Moving Forward

Although the NAMIMO reduced the error between the desired and measured radiator and engine temperatures, it was still not the most efficient when it came to electrical efficiency. The NAB controller consumed the least amount of power, perhaps aspects of this strategy could be integrated into NAMIMO in order to bring together the best of each controller. Several tests were conducted in order to compare the different controller, but test 7 was the only test conducted that varied the coolant flow rate along with the valve. This pump-valve actuation resulted in a decrease in the amount of power consumed by the pump. The pump makes up 15% of the total power consumption in the system, which is not nearly as much as the fans (85%) however it is a improvement nonetheless. Further research can be put into this controller strategy, and ultimately every component in the system will be regulated to operate as little as necessary to maintain the desired temperature and still be robust enough to handle external disturbances.

Conclusion

Moving away from fans that regulate the cooling air through the radiator as a function of crank speed is inefficient and an unneeded external load on the engine. Removing this load will improve the fuel efficiency of the engine alone, but this paper proposes a new method of regulating the engines coolant temperature by way of a smart valve. The valve replaces the classical thermostat that utilized wax as an actuator. Now complex controller such as NAMIMO can control the valves position and an array of radiator fans to ensure the optimal desired temperature is set for the engine. The most impressive takeaway from this approach is its robustness to disturbances, such as sudden increases in heat load from the engine and the effect of ram air on the radiator. Electrical power can be regulated to the fans as needed in order to conserve the total energy consumed by the system. Altogether the NAMIMO outperformed the NAB and SF control methods in almost all respects. Improvements can still be made by incorporating the electric coolant pump as a variable to adjust the mass flow of coolant in the system as well. In this case the power consumed by the fans could be reduced in addition to the pump operating more efficiently. The challenge will be maintaining the same robustness with respect to disturbances and still maintaining a sufficiently low error.

References

[1] T. T. Wang and J. Wagner, “Advanced Engine Cooling System Subjected to Ram Air Effect—Nonlinear Adaptive Multiple Input and Multiple Output (NAMIMO) Control,” in IEEE Transactions on Vehicular Technology, vol. 66, no. 9, pp. 7730-7740, Sept. 2017.