Operation

The logger flags5 were used to control programme flow. For each reactor, one flag was manually raised only during the compost growth phase. The second flag was used by the logger to indicate whether the reactor-air set temperature needed to be adjusted

(downwards during the growth phase – upwards during the decline phase). Three time frequencies were used in the control routine (Table 4. 1).

The rate at which the logger raised the reactor air temperature via heating resistors was determined by a manually inputted „increment‟. Activated on the 5 minute frequency meant the temperature could change by 12*increments per hour. An increment which closely matched the change in compost rate would minimise any non-steady state conditions and measure a more accurate signal. The magnitude of the increment decreased from a maximum immediately after peak composting, and only required infrequent altering after a couple of weeks of composting. A negative increment was input during the growth phase.

Table 4. 1– Logger control routines

Time interval _{DECLINE PHASE}

Flag low

Additional/different actions during the

GROWTH PHASE

Flag high

5 Second - Measure Temperatures

- Switch coil on/off

- Switch reactor-air heater on/off

Same

5 Minute - If air pump count <0; turn on air pump; calculate new delay.

- Else take 1 off delay.

- Raise reactor-air set temperature if second flag high.

Halve air pump delay.

Lower reactor-air set temperature if flag high.

30 Minute - Data to final storage

- Maintain coil contribution between 0.2 & 0.35 of total watts, by raising/lowering second flag.

- Lower reactor-air set temp if re-warming occurs in the compost.

Second flag raising is based on coil „on‟ count enabling a two stage response.

On occasions the composting rate increased again some time after the growth phase. A small amount of reheating could be accommodated by less energy being supplied to the coil, but when the coil contribution became too small the risk of the compost overheating increased. To prevent the compost overheating a control routine was activated when the proportion of heat from the coil was less than 12 % (compost proportion > 0.88). In this mode, a small decrease in reactor air set temperature occurred each 30 minute interval when this routine was triggered.

5.2.1.1 Instrumental Resolution

Platinum resistance temperature sensors vary their resistance by 0.385 Ω degree -1

. Thus, when connected to a CR10 logger with a resolution of 1 part in 7500 (of full scale range), it can be determined that the resolution of a single measurement is limited by the logger to 0.04 °C (± 0.02 °C).

This resolution can be further enhanced by averaging a number of readings. The point at which the resolution of the platinum sensors can be matched by the resolution of the logger final data storage (5 digits for high resolution output, or 3 decimal places for reactors whose set point is above 10 °C), is 0.04/0.001 or an average of 40 readings.

Thermocouple resolution is also limited by the logger resolution. At the 2.5 mV range the resolution is ± 0.33 µV. Copper-constantan thermocouple calibration is 0.042 mV.°C-

5

A boolean type software function which could be raised or lowered either manually or programmatically. They are used to control programme flow.

1

giving a resolution of 0.33 ÷ 42 = 0.008 °C. Only 8 readings are needed for the sensor resolution to match the logger resolution.

Variations in readings resulting from variation in the logger supply volts (caused, in part, by the current needed for the control ports), were detectable. The actual resolution of the final data would be less than the 0.001 °C absolute resolution that is possible.

5.2.1.2 Sensitivity

If the instrumental/ logger resolution is 0.001 °C and the UA of the reactors is 0.11 W.C-1, then the minimum watts that can be detected is:

0.11 x 0.001 = ± 1.1 x 10-4 W = ± 0.396 J.hr-1 For a 3 litre reactor, the sensitivity is:

0.396/3 = 0.132 J.hr-1.L-1

This compares with the absolute sensitivity of Luong and Volesky‟s (1983) thermal flux calorimeter of 0.042 J.hr-1 for a 10 cm3 sample: a sensitivity of 4.2 J.hr-1.L-1

By comparison, if the input air were 16 °C and 70% RH (enthalpy = 35 kJ/kg) and if 1.5 L of this air (FAS = 50%; specific volume 0.84 m3 kg-1) were humidified to 100% (enthalpy = 45 kJ kg-1) then 18 joules would be required, or 6 J L-1 for the 3 L reactor. With a sensitivity of 0.132 J.hr-1.L-1 the energy needed for humidifying this air (registering as a drop in the composting rate) would be detectable in 1.3 minutes. This effect was detected in the compost (it required 10-15 minutes to manifest), and was used to determine

optimum aeration (Section 5.2.2.2).

At this level of absolute sensitivity, the time required for quasi steady-state heat flows to manifest and the impossibility of such an exact steady-state, with heat output constantly changing, would limit the actual sensitivity of the reactors.

5.2.1.3 Stabilisation Time

With high resolution temperature sensors the time taken for stabilisation of the reactor increases. This arises from several sources:

 As stabilisation conditions are reached, the proportion of the heat flow going into storage decreases as a proportion of the total heat flow. The stabilisation curve is asymptotic and very flat as the resolution scale is approached.

 The logger control programme used the current reading of the Pt100 sensor (which had a resolution of 0.04 °C) to control the heater „on‟ switch. At stabilisation temperature the switching of the power to the heater was in part due to randomness due to the sensor reading. Statistically, at set

temperature the heaters would turn on 50% of the time (half of the readings would register above set temperature while the other half would register below the set temperature), while it would only be ± 0.02 °C either side of set temperature that the heaters would be beyond the randomness and switch reliably. For this reason the compost temperatures of trials were typically 0.02 °C above the set temperature, and it was often noted that the

temperature would drift upwards over a period of several days.

 At trial start-up the compost coil was run at 24 Volts to put 4 times more energy into the compost than the usual 12 V. This amount of power meant the surface of the reactor reached operating temperature some hours before the core of the compost pile reached operating temperature. Some of the energy would be going into storage (as sensible heat). Adjusting for this energy going into storage was not accounted for in reactor calculations. This substantially reduced the reliability of the composting rate

measurements over the first few hours of the growth phase. Bringing the compost mixture to operating temperature before loading into the reactors would enable stabilisation temperature to be reached much faster and improve the accuracy of the first few hours of data.

5.2.1.4 Logger Final Storage Frequency

A 30 minute final storage frequency was chosen and fitted the experimental purposes very well (the two cold temperature trials used a 1 hour frequency). Factors considered in this were:

 Identifiability, that is sufficient measurements to be able to determine the parameters. Fast fraction rate constants are largely degraded in a couple of days. Thus if 40 measurements are required to identify a parameter, then a frequency of less than 1 hour is needed.

 Data file size (particularly with longer trials).

 Clean signal – the trends over time were clear but the variability between measurements was minimal.

A shorter final storage frequency could be used for specific tasks, such as:

 Finer detail during the growth phase.

R1 r2

L

r1