• No results found

Chapter 5 Design Implementation Lessons Learned

5.1 Controls Deployment

There were several challenges that faced the controls team during the design and deployment stage. During the development stage, data needed from suppliers to develop the plant models was difficult to get. Within the plant models, the team was able to estimate the physics of the model based on papers published on the matter; for example how a battery would react during discharging. The lack of data with plant models arise with the plant controllers. The motor suppliers, engine developers and battery suppliers did not supply sufficient information on how their controllers would behave. Generally, suppliers would provide a list of CAN message. With this and past experience UWAFT would have to estimate the control logic. Once the device was bench tested (tested in isolation outside of the vehicle), the team was able to validate the estimated control logic. This process was generally time consuming and if estimations were incorrect, lead to major issues. One such example was the speed control mode within the generator controller. The CAN message list stated the generator could be operated in speed control; hence the entire generator control code was designed around this concept. Once tested on the bench, it was realized speed control did not work as intended. The control code had to be revamped to allow for torque control.

Another example of suppliers not provided sufficient information includes mimicking the GM CAN bus. Due to the removal of the engine and transmission, signals normally sent by these devices needed to be provided to the rest of the vehicle. Again, only the CAN list was provided; the team had to guess the control logic and sequencing when the vehicle was bench tested. The team also had difficulty detecting the gear shifter position. In the stock vehicle, a cable connects the gear shifter to the gearbox, which in turn measures the position of the gear shifter. With the stock gearbox removed, the team needed to find a new way to measure the gear shifter position; the team opted for a sliding potentiometer. A 12 V supply signal was provided to the potentiometer and the change in voltage

coordinated to gearshift position, which was sensed by the supervisory controller. The park-lock system, which does not allow for the driver to shift out of park unless the car is turned on and the brake pedal depressed, remained operational without the gearbox.

An issue that was not foreseen during the design phase was the current limitation on the controller. The controller’ current output is limited to a maximum of 250 mA. The initial design called for the controller to activate relays for fans, pumps, and device wake-ups. The current required to drive these relays was above the 250 mA limit. The team reverted thus had to design a Darlington transistor which amplified a digital output signal from the controller. The team also investigated constructing a separate CAN controlled relay box, but the idea was discarded due to a lack of people able to design the device.

As mentioned in Section 4.4.2, updating the SIL/HIL models based on updated information from deployment of vehicle code is important. Up to date SIL/HIL model, control code can be updated in parallel and offline optimization can take place. Unfortunately for UWAFT keeping the SIL/HIL models up to date was challenging. The major issue due to this was licensing. The team was provided with unlimited MATLAB licenses, however there were only a limited number licenses for dSPACE controller software. This limitation meant only a handful of students could update the vehicle control code. To mitigate this issue in the future, UWAFT plans to designing a ‘wrapper’ for the control code. This allows for the kernel code to exist without any dSPACE block sets, allowing for anyone to update the control code. The dSPACE block sets can be added just before the code is deployed to an HIL or on the vehicle. This proposed ‘wrapper’ layout in shown in Figure 60 , an update to Figure 27

110

Diagnostics Hybrid Control Strategy Component Control Hybrid Control Torque Control Thermal Control Battery Engine Front Motor Rear Motor Generator Rear Park Lock

Charger 12 V DC/DC Custom Acutator From Vehicle To Vehicle Vehicle State (Start-up, shutdown, EV, Hybrid Mode) Battery Engine Front Motor Rear Motor Generator Rear Park Lock

Charger 12 V DC/DC Custom Acutator

Controller Code Kernel

Component Interaction Control Engine + Generator Battery + Motors + Generator + HV Accessories Front Motor + Rear

Motor Battery + Charger Rear Motor + Rear

Park Lock dSPACE

Blocksets

dSPACE Blocksets

Figure 60 - Updated UWAFT control code with dSPACE wrappers

111

Beyond the technical challenges of code deployment, human capital was major issue for controls development. Human capital is a broad term, which includes the amount of students, the skills they possess, their commitment to the project, and creativity towards solving problems. For UWAFT, attracting students issues was not an issue; attracting students with the correct skill set was a challenge. Student in software, electrical or computer engineering, which is the required skills for controls development, did not see an automotive based project has having relevance to their field. These groups of students saw automotive design as a mechanical engineering task. The reality is most new cars (specifically EVs, HEVs and PHEVs) contain almost 50 embedded systems that are all within the domain of electrical, computer and software engineering. It is also challenging for the automotive industry to compete with software companies, which typically offer higher pay, better benefits, and better job locations.

Most students working on controls development were mechanical engineering students who learned controls development through self study. This worked well, however mechanical students lacked motivation to learn a new skill set, and many found controls more abstract than mechanical design. Mechanical design is easy to see, touch and hold; controls code is abstract and difficult to explain. Moving forward, UWAFT is working to engage more electrical/computer/software engineering by showcasing the need for their skill sets within the industry. Thanks to the success of Tesla Motors in California and the start-up like approach to operation, UWAFT also hopes to use their achievement to attract more students to the automotive industry.