Overview

In order to meet our requirements of delivering enough power to the pressure vessel via heating from the solar collector, we had to make changes to both major components. In terms of the chamber, we need to find a better way of monitoring conditions inside the pressure cooker. In addition to measuring temperature, we want to measure the pressure to get more accurate and precise readings. The design of electronics and sensors is crucial to this goal, and we hope to increase precision by amplifying the signal and creating a sensible and efficient structure. In relation to the solar collector, we had to make considerable changes to come closer to meeting our requirements. The solar collector from the first iteration was not portable, but more importantly was not efficient at focus light due to artifacts on the Mylar surface during construction. The construction was also more involved than what we would have liked, especially considering how much the ultimate users of the autoclave would benefit greatly from easy assembly and ability to repair the system.

The two components were designed and constructed in parallel, in hopes of joining the two after considerable improvements were made on both ends.

Chamber

Three sensors are used to monitor the conditions inside the vessel:

• Pressure transducer from DigiKey (model MPXM2202AS-ND): A temperature compensated silicon pressure sensor, rated to 29 psi (absolute) and temperatures up to 125 °C. An absolute pressure transducer was chosen due to the challenges fixturing the sensor - it was easier to have the sensor sitting completely inside the pressure cooker instead of mounting in the lid.
• Thermocouple (K type) from Adafruit (http://www.adafruit.com/products/270)
• Resistance Temperature Detector (RTD): Operates on the principle that the electrical resistance of certain metals changes in a predictable manner with temperature.  Benefits: lower cost, potential for higher accuracy and repeatability than thermocouples.

The LM324 quad op amp, which operates with 5 volts, was used for non-inverting amplification to provide 10 volts for the pressure transducer and to amplify the outputs from the transducer and RTD.

-To power the transducer, 10 volts was achieved with two 120 Ω resistors.

-The transducer output was multiplied by a gain of approximately 100 using 15 kΩ and 150 Ω resistors.

-The RTD was used in series with a 47 kΩ resistor to make a voltage divider across 5 volts. The resulting output was amplified with a gain of approximately 20 using 1.5 kΩ and 120 Ω resistors

In this way, all sensors were powered directly from the Arduino Uno board. The Arduino remained connected to a computer for power and recording sensor output. More exact values for resistances were measured using a multimeter.

The three sensors are slotted through a hole drilled into the lid.  To maintain closure of the vessel during operation, the hole is sealed with "J-B Weld" steel-reinforced epoxy (http://www.jbweld.com/product/j-b-weld/).  This epoxy can withstand temperatures of up to 550ºF (288ºC) when fully cured and has a strength of 3960 psi.  The resistance of the epoxy was checked with a multimeter (and was found to be very high) to ensure that the epoxy would not bridge any of the electrical connections to the sensors.

The image below shows an earlier test with only the thermocouple, mounted using a different epoxy. While epoxies should be unchanged by steam at 121 °C, this formulation yellowed after use, so the J-B weld product was used for later tests.

Because the solar collector was not yet ready for use with the vessel, the first tests of the chamber were done using a stove. 

With no lid, we measured the time required to heat a known volume of water (3 cups and 6 cups) over a fixed temperature range (30-45ºC) at burner settings 2 and 4 in order to estimate the power output of the stove at these settings. Although power is clearly lost to the surroundings, we estimated a lower limit for the power provided by the stove using the standard specific heat of water.

We estimated that the stove output 0.3 kW at setting 2 and 0.6 kW at setting 4.  These figures are both significantly greater than our target solar collector performance of 0.25 kW.

In light of this, we wanted to see if the pressure cooker, when closed and filled with water, would reach the desired temperature and pressure (121ºC and 15 psig) on the lowest burner setting.  To attempt to reduce the amount of steam escaping from the chamber, as well as heat transfer through the walls, we covered the sides and lid of the pressure cooker with Owens Corning fiberglass insulation (2" thick), leaving the area of the lid with the wires uncovered and cutting a hole for the pressure regulator valve and yellow pressure indicator.  However, the insulation did not actually prevent the outflow of steam, as evidenced by the fact that it became wet during testing.  In fact, with the stove on the low heat setting, the it took longer for the water to heat from room temperature to boiling with insulation (17 min) than without insulation (13 min).  This may be due to imprecision in the burner.

It was observed that a greater volume of water required less power input for the vessel to reach adequate temperature and pressure (121ºC and 15 psig).  When the vessel was filled with 6 cups of water, a burner setting of at least 3 was needed.  For lower burner settings, the temperature abruptly leveled off around 100ºC, the boiling point of water.  This persisted for more than 10 minutes, and the yellow pressure indicator never popped up.  On the other hand, when the vessel was filled with the maximum recommended amount of water (2/3 full or about 11 cups), it was able to reach the desired temperature/pressure at a minimum burner setting of 2.  The pressure cooker is not completely sealed; steam was observed to vent from the sides of the lid handle and also the pressure regulator valve even before the target pressure was reached.  This may explain why the vessel could not become pressurized below a certain burner setting; the rate of heat input may not have been enough to compensate for the mass loss due to the escaping steam (plus conductive losses).

We concluded that the pressure cooker was not a suitable for future iterations - too much power was lost at lower temperatures, and the maximum power input seemed to be determined by what was needed to get the pressure indicator to close at 100 ºC. We expect to achieve greater efficiency with a vessel that does not vent steam at pressures below 15 psi.

The pressure transducer gave an output of 20 mV when powered in the lab, corresponding to atmospheric pressure. However, the transducer did not work during testing with the pressure vessel - it gave 0 mV consistently. It's possible that the transducer failed from an excessive pressure when testing the fixture for leaks. It was decided that the pressure limit for the selected transducer was too low - shocks during setup or operation could significantly exceed the pressure limit.

Lastly, while the RTD measurements varied upon heating the vessel , we were not confident that we could get greater precision than with the thermocouple. The measured resistances were not as consistent as thermocouple readings. This could be from imprecise analog readings by the microcontroller, changes in contact resistances where there was an unsoldered mechanical connection in the circuit, or from the lack of regulation of the applied voltage.

Solar Collector

The most important aspect under consideration for the second iteration of the solar collector is the geometry, for the geometry of the Mylar dictates the amount of light collected by the collector and focused on the vessel. The previous prototype had led us to look for simpler geometries that may not be as efficient as focusing light but were still capable of delivering the right amount of power to the chamber and were easier to construct and repair. We considered a parabolic trough that would allow us to make a line focus along the width of the collector, but ultimately decided against this design because of the severe geometric constraints. Instead, we decided on a catenary for the geometry of this prototype, which is created by simply letting the Mylar hang by its own weight. 

Before the construction of the new collector, we ran some calculations to determine efficiency losses from choosing this design, especially because there is no point focus for a catenary. Preliminary calculations led to to believe that by maintaining a 4 to 1 aspect ratio (width of parabola to depth), we could successfully replicate the focus of a parabola. The graph below (from http://thoxbui.com/) uses ray analysis to determine the focal points of a catenary (left) versus the focal point of a parabola (right).

Based on the above geometry and accounting for shadowing effects of the pressure cooker, we determined a 80% efficiency which is more than enough for operation of the autoclave at a height of 0.83 ft from the base of the collector. The results from our ray and efficiency analyses can be seen in the plot below.

The construction was rather straight forward: a portable stand and a 'scroll' of Mylar that would hang on the stand, with the weight of the Mylar creating the catenary shape. The final assembled and packed solar collector can be seen below.

Although we had designed the geometry to have a focal point exactly 1 foot above the midpoint of catenary, it was important that we verify this experimentally. The solar collector was set outdoor, and a piece of paper was used to determine the location of the focus locus (since it is not a single point for a catenary). The image below shows a paper test to find the focus. Unfortunately the collector was almost useless in terms of other types of testing because it could not withstand the wind, and any small air movements led to exaggerated wrinkling and movement of the solar collector.

While we had hoped to measure the light intensity at the focus relative to the incoming light intensity, these measurements were not accurate for this model due to other complications, including saturation of the light sensor and wind problems. The next prototype will address problems faced in this one, to get closer to meeting our requirements. It is noteworthy to mention that the solar collector and stand were easily portable since the entire assembly was light and no greater than 3 ft in any dimension.