Wednesday, May 11, 2011

Future Work

This was a very relevant project to today’s society. Although biodiesel is not currently competitive economically (especially due to all the costs involved in making the technology needed to produce it), this would be a great project to continue. In the future, groups should ensure that their catalyst is functioning properly before using it. The method for testing the catalyst is very complex and would have to be done by an outside source, but it would be worth it to know that the catalyst is not poisoned. Also, if there were still problems in the lab, the use of ultrasonics for dynamic mixing would be a good possibility. Loyola has access to an ultrasonic cleaner which could agitate the reaction mechanically in order to promote mixing and enhance results if needed.

Additionally, it would be essential to do further research regarding the CFD-ACE+ software to ensure that everything is scaled properly. We were satisfied with the results that we obtained from the simulations; however, there was no way of telling for sure that those results were accurate. Therefore, once the laboratory tests are successful, comparisons should be made between the simulations and the experimental results. At least five simulations should be run with a given set of parameters and five laboratory tests should be run with those same parameters to test for compatibility. Once results begin to agree, new values may be acquired to optimize a commercial scale design. Lastly, to help make this project more cost-effective and beneficial, it would be a good idea to look into less expensive feed stocks, or even waste materials from kitchens. Often times restaurants dispose of a lot of cooking oils, but if a way could be figured out to purify these waste oils and then use them to produce biodiesel, that would be a great step in the right direction. Clearly, there are many ways to improve upon this project and we would like to see it carried out by future groups.

Project Management

Above is a detailed layout of the way we approached the semester. We set timelines for different tasks and assigned specific jobs to each member to share the work load evenly. By sticking to this, we were ultimately able to produce an effective presentation and written report, despite failing in the lab.

Project Budget

The above table shows that the total cost for the entire project was $96.21 and this project had a $1,000 budget. Clearly, the group was very frugal with its money. Because the total cost of the project would have amounted to $3,792.19 without the aid of Loyola University and its donated items, the group was very diligent in obtaining items for free when at all possible. The group proved that making use of connections and all possible resources is a viable way to reduce the cost of a potentially expensive project.

Solar Energy

Here is a photo of the solar array that we used in our lab tests.

The Green Machine makes use of photovoltaics for solar powered energy. The solar array will be used to power both the pump and the immersion heater of the Green Machine’s design. For the current solar array being used by the group, as shown above, each solar cell is rated at 1000 W/m2. In the laboratory setup, the solar cells are wired to a SunGuard Solar Controller. This controller delivers the converted solar energy to a 12V DC lead-acid battery, which connects to an AC-DC Powerverter. Using the Powerverter's output of 115V AC, the Green Machine's immersion bath and syringe pump system is powered. The total solar cell area on this array is 0.147 m2, which only produces 147 W of power. Therefore, calculations were performed to determine the necessary power requirements and solar cell area for the commercial scale-up design. The entire system will require 106.1 kW of power, which means a solar panel area of about 108 m2 is needed. It should be noted that this calculation of solar area adds about 2% extra area than what is specifically needed, as was recommended by research done on solar panels, to ensure that there is ample power available.

Software Simulations

3D flow modeling was used in the CFD-ACE+ program. The following figure shows the unit cell pattern used in the software, which captures the presence of the triangular obstructions within the channel.

Figure 6: Unit cell with triangular obstruction pattern

(Note: A microreactor channel would simply consist of repeating units of this cell)

Using the different values that were determined from the spreadsheet analysis, the group was able to run many different simulations. Table 2 below shows the highlights of a convergence study that was performed to ensure that the design could meet the specification of 100% conversion to biodiesel. Several different tests were run at various molar ratios with excess methanol to drive the reaction and alter the flow rates of the reactants.

Table 2: Convergence study of biodiesel conversion

As the results indicate, a molar ratio of 25:1 for methanol to soy oil produced 100% biodiesel conversion. Both the methanol and soy oil would be completely consumed and there would only be biodiesel remaining. As shown in Figure 7 below, the biodiesel concentration gradually increases as the reactants move down the channel. When the biodiesel is at the end of the channel, the hot pink color indicates that full conversion has occurred, as evidenced by the fact that it is at the maximum on the color coded scale shown below the model.

Figure 7: Biodiesel concentration with 100% conversion

When too little methanol was used in the simulations, soy oil would remain in the channel. However, if too much methanol was used, excess methanol would remain in the channel. Therefore, it was clear that a 25:1 molar ratio was needed to meet specifications. Once this number was finally determined, all the corresponding calculations were carried out using the group's Excel spreadsheet. The actual numbers yielded is discussed in the Final Design post.

Theory for Design

The CFD-ACE+ flow modeling software served as the backbone for the group’s system design. The theory behind this software involves the conservation of mass, conservation of momentum and conservation of energy equations. Together these equations form a set of coupled, nonlinear differential equations. Because these cannot feasibly be determined analytically, this program finds computer-based solutions governing these equations for fluid flow. Also, because the group did not have success in the laboratory, this software became an invaluable resource and guided the process of design selection.

In order to get the software functioning properly, a 3-Dimensional model of the microreactor used in the lab was created. Next, various properties such as density, viscosity, molar mass and chemical formulas of the fluids (soy oil, methanol and biodiesel) were defined. Then, the chemical reaction of transesterification spurred by the NiO catalyst was added to the program. That is, a reaction was added at all surfaces of the microreactor channel in which a molar ratio of 3:1 for methanol to soy oil produced 3 moles of biodiesel. As research from the University of Tennessee indicates, the kinetics of biodiesel formation (soy oil reaction) catalyzed by NiO is first order with a rate constant of k = 0.0002 per second. This value is part of the following relationship:

[soy oil at t] = [soy oil initial] exp(-kt)

where [] means concentration of, and t is time. This rate constant was determined at 80˚C, which is precisely the temperature that the Green Machine operates at. Therefore, this constant was incorporated into the software to determine the rate at which biodiesel was formed in the channel. One other key parameter used in the software is diffusivity. While it is known that soy oil and methanol are very immiscible, meaning they do not readily mix, there is no concrete number for their diffusivity. However, for liquid-liquid diffusivity at elevated temperatures such as 80˚C, a diffusivity of 10-9 N s/m2 is a reasonable assumption. Therefore, this is the value that was used in the CFD-ACE+ software.

Another key tool that aided the software was the use of spreadsheet analysis in Excel. By setting up one spreadsheet, the group was able to manipulate variables and run different simulations in order to optimize results. Clearly, optimization is key to the Green Machine’s design. Essentially what is being optimized is the number of microreactors needed in the system. The goal here is to use as few microreactors as possible to produce 1 gallon of biodiesel in 24 hours, while keeping the pressure drop in each reactor under 1,300 Pa. Consequently, there is an optimal microreactor channel length, residence time, velocity and flow rate that must be found first in order to get the optimal number of microreactors needed. Therefore, the two most important variables to manipulate in the Excel spreadsheet were the length of the microreactor channel and the residence time of the reactants in the channel. By changing these two variables, the group was able to alter the flow rates and velocity of the reactants, and ultimately, the number of microreactors. As a first cut, the group chose a 5 minute residence time to provide ample time for mixing, and a microreactor channel length of 3.5 cm because it was the length of the reactor being used in the lab. Starting with these assumed values, a spreadsheet was set up to calculate flow rates and velocity according to the following two equations:

V = L/t

Q = VA

where t is the residence time, L is the length of the microreactor channel, Q is the flow rate, V is the velocity of the reactants, and A is the cross-sectional area of the channel. As shown, the flow rate is directly related the velocity, and it was essential to know the velocity because the software required a velocity input in order to run simulations. It should be noted that the Q in the above equation represents the total flow rate which is simply the sum of the flow rate of methanol and that of soy oil. In order to get each individual flow rate, it was essential to make use of the chemistry and molar ratios of the two fluids.

The optimization of the Green Machine design was an iterative process. Therefore, the group calculated initial values to fill in an Excel spreadsheet, ran a simulation, analyzed the results, altered the parameters and then repeated the process many times. With the initial values that were chosen of a 5 minute residence time and 3.5 cm microreactor channel, the number of microreactors needed was 15,000. This number was far above the performance specification of 1,000 microreactors or less, and hence, many iterations were performed until the group finally arrived at its final value of 750 microreactors.

Tuesday, May 10, 2011

Root Cause Failure Analysis: Part 2

Initially, the difference in densities of the reactants was believed to be the cause of the squishing effect. To test this theory, CFD-ACE+ software was used to model flow conditions before, at, and after the channel junction. The results appear below in the figure.

Figure 2: Reactant concentration as a function of distance into channel

In the figure above, the pink color represents concentrated soy oil, while the blue color indicates concentrated methanol. A cursory glance at the channel junction confirms that the squishing effect indeed takes place. However, the simulation shows that mixing should occur after about ten unit cells into the mixing channel. This is indicated by a gradual color change from pink and blue to orange. Stronger hues of orange seen further along in the channel indicate more thorough mixing. Therefore, the simulation confirmed suspicions of squishing but also proved it had little impact on the group’s failure to synthesize biodiesel. Furthermore, it was determined that the difference in fluid viscosities was the driving force behind the squishing effect, as opposed to the difference in densities, as originally theorized.

Next, non-ideal temperature was investigated as a possible failure mode. As a general rule of thumb, most chemical reactions increase in rate as temperature increases. In light of this, relatively high reaction temperatures in laboratory experimentations were sought. Tests conducted in the first semester demonstrated a hesitance to increase reaction temperature beyond 60°C due to the boiling point of methanol, which is 62.5°C. Second semester tests utilized reaction temperatures of 70°C and 80°C. The latter temperature was especially important for two reasons. First, the rate constant of the reaction was determined experimentally at this temperature and subsequently used in the software simulation (as will be discussed in detail in the Therory for Design section). Second, full conversion of biodiesel at 80°C was achieved in batch testing at the University of Tennessee at Chattanooga (UTC). Thin Layer Chromatography (TLC) analysis indicated no presence of biodiesel in the product at both elevated temperatures. To ensure no conversion whatsoever, a test was run at 80°C and its product sent in two samples to UTC for nuclear magnetic resonance (NMR) testing. These tests also confirmed that no conversion had taken place. Additional information on TLC analysis appears in Appendix C.

Finally, the nickel oxide (NiO) catalyst was investigated as a potential cause of failure. During experimentation, the group had to operate using several critical assumptions. First, it was assumed that the catalyst was reusable, that is, no nickel oxide was consumed in any reaction in the mixing channel. This was a reasonable assumption to make considering catalysts do not participate in their reactions by definition. Second, the group assumed that the catalyst was bonded permanently to the mixing channel walls and was not removed during flow. Finally, it was assumed that the catalyst was indeed active during flow through the channel. The basis for this was a study done by UTC in which nickel oxide was shown to be a viable catalyst for biodiesel synthesis. Comparing the forms of nickel oxide introduced to the reactants in UTC’s batch testing and in the group’s microreactor gave some critical insight in that the former method utilized nickel oxide nanopowder, while the latter used a sputtered catalyst. It was theorized that the force of the nickel oxide against the channel walls during sputtering had deactivated the catalyst, thereby destroying its catalytic properties.

After ruling out bad mixing and non-ideal temperature as possible failure modes, it was determined that deactivation of the NiO catalyst was the root cause of failure.