Initial Observations:

• As observed previously, P_reactor and P_filter empirically show a proportional relationship. This can be used as a part of a tuning method for the filter media, and to monitor packing of the filter bed
• Muffler and pyrocoil show clear time constants (exponential decay/increase of temp with time). Model fitting can be used on these to understand system dynamics. Time constants can also be found for reactor startup (although noisier) in order to understand the starting time required (and as a way to quantifiable number for optimizing start up times).
• Muffler has the shortest time constant ("t" in units of time, t is equal to time taken to reach 63% of final temperature). Pyro-input has longer t, and pyro-output has longest (as expected: the pyro-coil is bringing a mass of wood up to temperature).
• While the engine was starting/running with wood pellets in the pyro-coil (not auger fed), the relationship between T_pyro_in and T_muffler fit a model close to: T_pyro_in = 0.88 * T_muffler. This was without some insulation around the lower flex-connector. Mean value while running (T_pyro_in/T_muffler) was ~0.91 (e.g. at 600°C exhaust, pyrocoil was running at ~550°C).
• Once reaching equilibrium, exhaust temperatures stayed at (or just above) 600°C.
• While running, pyrocoil output temperatures were 54% of input gas temperatures (e.g. at ~550°C incoming temperatures, outgoing gas was ~300°C)
• Manual TC measurement of combustion temperatures at flare nozzle while running on wood pellets showed temperatures in excess of 1100°C.

Time Series Plot

Gas Composition

Notes:

The precision of the CO measurement can't measure CO concentrations at PPM levels (e.g. in exhaust). According to this Iowa State FAQ, CO levels in car exhaust before a catalytic convertor reaches temperature can excede 8%.

Determination of Time Constants

The pyro-coil and muffler show obvious simple dynamics. See the wikipedia page on time constants in modelling of thermal systems.

Engine On (warming) (datalog time from 3360 to 4800 s)

Engine Off (cooling) (datalog time from 5760 to 7320 s) 

Using the same naming conventions as wikipedia, where:

V(t) is the temperature over time

V_0 is the initial temperature [°C]

tau is the time constant (63% of the change occurs over this time) [in seconds]

A*tau is the final temperature (V_infinity) [°C]

Comments:

• The initial temperature is very dependent on when one considers the engine started or the engine off, so is relatively meaningless.
• The time constant tau modeled the exponential decay (or increase) such that: 63% of final temperature is reached in 1*tau seconds, 80% in 2*tau s, 93% in 3*tau seconds.
• Thus, on start up, it takes the muffler about 10 minutes to reach 90% of its final temperature. The pyrocoil input takes about 11 minutes to reach >90% of its final temperature (more cumulative thermal mass). It takes the pyrocoil 26 minutes to achieve >90% of its final temperature (showing much greater thermal mass, suggesting the biomass is acting as a load).
• The final temperature of the muffler on shutdown is 0°C (outside ambient temps during the run were close to freezing), however the GCU doesn't cold point compensate the TC measurements.
• Interestingly, the pyro-coil in and out final temperatures on shutdown both end exactly at 85°C. This could show the heat output from the pyrolyzing biomass and/or its large thermal mass, note that once the engine is shutdown, the primary thermal transfer is via conduction (and given the ports are outside the pyrocoil and exposed to ambient temps, the final temps inside the pyrocoil might be 2-3x the final pyro in/out temps).

R code:

> t <- seq(0,7320-5760)

> m <- nls(T6_muffler[5760:7320] ~ V_0*exp(-t/tau) + A*tau*(1-exp(-t/tau)) ,start=list(A=3,tau=100,V_0=600))

> m