You are here

Cookstove Design

Alex Wohlgemuth, a senior in mechanical engineering, has been conducting research under the guidance of his faculty advisor, Sandip Mazumder, and a local mechanical engineer, Dale Andreatta. Dr. Andreatta works with developing world cook stoves and similar projects in his spare time, and saw the benefits of modeling certain scenarios in a multi-physics solver program such as CFD.

Creating more efficient cookstoves has numerous benefits for those in developing nations. Less wood is required, meaning the local ecosystem benefits, and less smoke is produced, which benefits the health of those cooking. Furthermore, more efficient stoves mean shorter cooking times, so those in charge of cooking now have more time for other activities, such as spending time with their family or other personal endeavors. With this in mind, the following is a fairly extensive overview of the research process.

The first step in conducting research involves creating a 2-D (or 3-D if necessary) model of the object of interest in a program called CFD-GEOM. Grids of varying size (more lines and smaller spaces create a denser mesh, and higher accuracy) are used to fill the areas where interactions of interest take place. In GEOM, the user can identify areas as regions where material can enter or exit the problem. Other areas are designated as walls – real-life boundaries to the problem (such as the bottom or side of a stove). Once this model is complete, the program CFD-ACE is used to set up the conditions of the problem in question. For instance, at the inlet, a substance designated ‘cook stove gas’ was modeled with properties identical to air. For a particular scenario, it was given specific conditions such as temperature and velocity (e.g., 773 K, 1.18 m/s). Once all the conditions of the problem have been specified, the problem is submitted to the solver. These results are displayed in CFD-VIEW, where a great deal of data can be recorded and analyzed. Continuing this example, one can determine the temperature and velocity profile of the gas as it enters the region of interest, comes into contact with the bottom of the cook stove, and, due to natural convection, continues along the side of the pot.

Furthermore, one can determine the heat flux transferred to the bottom and sides of the pot – this was the focus of the initial testing. Dr. Andreatta also proposed the possibility of having two concentric rings of air entering beneath the pot – one at a significantly higher temperature than the other (both rings had proportionate velocity values). Conditions were calculated so that in both scenarios – that in which the air entered at two different temperature and velocity values, and that in which the air entered at one temperature and velocity – the energy required was the same. However, the flux to the bottom of the pot was greater when the air entered at varying temperature and velocity values.

Other scenarios have been investigated, mainly variations on the addition of a skirt to the pot. The theory behind a skirt is that it improves heat transfer to the pot in two ways: by increasing convective transfer (transfer from a fluid to a solid) by pushing the gas closer to the pot, and by increasing radiative transfer by absorbing heat from the gas, which in turn creates a temperature differential which promotes radiative flux from the skirt to the pot. Numerous tests were run taking into account various factors. The first two, turbulence and absorption coefficient, were intended to make the computational model as similar as possible to the real-world situation. The last factor, the skirt, was the main focus of the testing. Here is an overview of the factors investigated:

Turbulence of the flow

(no turbulence – flow is in layers, or varying degrees of turbulent intensity, representing internal mixing of the fluid, leading to variations in velocity of the fluid)

The flow was modeled with no turbulence, as well as with intensities of 2.5%, 5%, 10%, 20% and 30%. Differing values in turbulence didn’t affect the power input of the system, but did affect where that power went. Greater turbulence meant that more convective heat transfer occurred on the bottom of the pot. A turbulent intensity of 10% appeared to match real-world tests most closely. Again, a base case of laminar flow was run to provide a comparison. Turbulent flow almost always increased heat transfer over laminar flow.

Absorption coefficient of the gas

(a property of the gas that represents the sootiness of the flame and the extent to which it participates in radiative transfer)

The gas was given several different values of absorption coefficient – initially, 0, 1, and 5, and later, values of 3 and 4. A value of 0 corresponds to a non-participating medium, and the remaining values represent different volume fractions of soot in the gas. A gas with an AC of 5 is considered very sooty, and the most radiative transfer occurred in this case. A value of 3 was found to match the real-world tests the closest. Tests were also run ignoring any radiative effects as a way of checking the accuracy of the model.

Distance between the skirt and the pot

(too far away, and the gas is not pressed closer to the pot, nor is there any significant radiative gain; too close, and the skirt obstructs the flow, and convective transfer lost offsets any gain by radiative transfer)

A perfectly insulated skirt, about 2.5 mm thick, was placed at several distances from the pot – 25, 15, 10, 7 and 5 mm. 10 mm ended up being the ideal case, where the flow was restricted enough to ensure maximum convective transfer, as well as providing significant radiative gain. These were all matched against the case without a skirt to provide a gauge of improvement over this base case.

Recently, the skirt was re-modeled as steel, instead of perfectly insulated, to again better match the base case. Since steel has a very high thermal conductivity, meaning energy is easily transferred through it, there was significantly less radiative transfer coming from the skirt to the pot. In an insulated case, no energy can get through, so it is forced to reflect back to the pot. Furthermore, the mesh was modified to allow the skirt to be moved up and down – in the mesh, this meant breaking it up into half-inch tall segments, which could be treated as fluid or a skirt, depending on the objectives of the tests at hand. An insulated skirt increases heat transfer by about 50%, whereas the steel skirt increases it by 25%. A next step is to investigate the benefit gained by adding a highly insulated material, such as glass wool, on the outside of the skirt.

Dale Andreatta recently presented these results at an ETHOS conference in Washington State. He made a powerpoint with this data, which can be downloaded on this page soon.