Development of a numerical model for power module cooling infrastructures
Prodrive technologies is a developer and manufacturer of world-class electronics. It comprises over 1300 employees and is one of the fastest growing companies in Europe. Our organization is based on two groups: Development and Operations. The Development department is oriented at developing first-rate electronic, mechanic and software solutions. The operations department is responsible for production, assembly, testing and life-cycle-management of electronic products and systems in the range of 1 to more than a million pieces per year.
Within Prodrive, numerous products incorporate power modules and discrete semiconductor devices. These modules package power semiconductors and are often off-the-shelf products which are not typically optimized for thermal performance. In line with Prodrive’s ‘vertical integration’ way of working, we have started developing our own customised power module designs.
The power semiconductors generally dissipate a significant amount of heat due to switching and electrical conduction losses. The heat dissipation leads to the materials in the package heating up. The problem is increased by the pulsation of power leading to the power dissipation in the transistors to vary, resulting in significant thermal cycles. Due to the variation in thermal expansion coefficient of the materials in the module, changes in temperature induce stress both in and between the various layers. Hence, thermal cycling is the main reason that power modules will fail in time. In many products the power device is the limiting component in terms of lifetime and reliability, improving the design or cooling of the module therefore has a direct impact on wider system performance.
There are multiple methods to optimize the transistor (referred to as die) packaging, for example by minimizing the thermal resistance between junction to coolant, matching the thermal expansion coefficients of the materials used in the package, or by optimizing the design electrically to decrease the electrical losses. These solutions can lead to higher lifetime or higher power density when using the same die, or to lower cost for the same performance by using lower rated dies.
When moving from Silicon (Si) to Silicon Carbide (SiC) or Gallium Nitride (GaN) technology the heat flux significantly increases as the same power is dissipated over a smaller surface area. The smaller surface area leads to a shift in the contribution of each material layer to the total thermal resistance from die to coolant.
Develop a numerical model for power module cooling infrastructures
- Develop a numerical model which is able to predict the thermal behaviour of a selected power device when mounted to:
- the most common coldplate within Prodrive, a Prodrive custom coldplate
- a pinfin baseplate (liquid cooled)
- Convection cooled extruded heatsink
- Verify the model experimentally (TBD: by FEM)
- Analyse the various cooling strategies and identify areas for future optimisation
- The numerical model should consider the following:
- Die size, material, losses
- Stack materials and geometry
- Cooling strategy employed
- The significance of flow rate, pressure loss, coolant type etc. during operation
- The influence of other power devices in the vicinity
Important note: all our graduation assignments are dependent on time. Please see them as an example of the many great things you can do at Prodrive Technologies. If you have the same passion for technology as we do, we always have an assignment for you!
Benefits as a student
Next to the benefits which are true for every Prodriver that are listed below, some special benefits must be highlighted for students. First of all you will be treated as a real Prodriver. Your opinion and ideas will be heard and treated as full. Next to this you will get the chance to work with colleagues with a very high level of knowledge in their field and get the chance to use our world class lab and production facilities for your graduation assignment.
Apply for Development of a numerical model for power module cooling infrastructures