Analysis of the Celani calorimeter Edmund Storms
12/16/12 The Celani calorimeter is the isoperibolic-type with glass being used as the thermal barrier. The temperature gradient across the barrier is measured at one location by making physical contact with a thermocouple on the inside and with another one on the outside of the glass. The device is calibrated by heating an inert wire within the enclosed gas. The wire being studied occupies a similar space and is heated separately. Such a design requires calibration be done using exactly the same conditions as the proposed measurement. The isoperobolic-type calorimeter relies on measuring a temperature gradient across a thermal barrier through which all energy passes. Stability requires the reference temperature be constant and the delta T represent the average at all points on the thermal barrier. The Celani design fails on both counts. The reference temperature (outside surface of the glass) is not constant and the inside temperature does not represent the average inside temperature because it is measured at one arbitrary location. Heat generated by the wire will create convection currents within the gas and these will heat the glass in uneven ways with respect location and time. In addition, these convection currents will not be the same when the gas is heated by the calibration wire and by the wire being studied. Therefore, determination of this potential error is the first important measurement. This calorimeter will have a delay between when the power is changed and when the temperatures become constant. Determining this time is an essential measurement. No temperature measurement has any value until this time has passed. Consequently, calibration must be taken as individual points, with the required delay between each one, first going up in temperature and then an equal number taken going down in temperature. These points should be plotted as applied power vs delta T to which a least squares equation of the form W = A + ∆T*B + ∆T^2*C is applied. This equation is then used to calculate any excess by subtracting the amount of applied power. Because loss by radiation will be the same during calibration and during the study of the Celani wire, it can be ignored. This sequence must be done several times under various gas compositions and pressures to determine the effect of these variables. Any lack of consistent behavior reveals the magnitude of the error in the measurement. Future use of the calorimeter to study a potentially active wire will be limited by the magnitude of this variation, with no excess heat being considered real if it lies within this error band. Consequently, this error band must be evaluated before the calorimeter is put to use and the magnitude reduced if possible. Until, the magnitude of this error band is determined no result using the Celani wire can be trusted. Errors can be expected from the following sources: 1. The outside temperature will fluctuate as a result of changes in room temperature and random convection currents around the calorimeter. 2. The inside temperature will fluctuate as the convention currents within the gas change with applied power and time. 3. The pressure will change as the amount of power is changed, which will change the amount of heat communicated to the glass. This change will not be reproducible by some unknown amount. 4. The calibration wire and the Celani wire will not produce the same convection currents, hence the calibration equation will not represent the behavior of the wire by some unknown amount. In general, the more places the ∆t is measured on the glass, the better. The more isolated the system is from random convection currents in the room, the better. Use of a fan blowing on the outside can reduce these effects. A basic problem exists in how this device is designed. The Celani wire will be heated to some unknown temperature during calibration by convection currents. The amount of excess power this temperature will produce is unknown. The Celani wire and the calibration wire will produce different convection currents, hence will heat the single spot where ∆T is measured differently. Consequently a valid calibration is not possible. The only hope is that the magnitude of the excess power is great enough to overwhelm these errors. So far, this has not been demonstrated.