Efficiency Enhancement in Dry Roasting Processes Using Hybrid Gas-Electric Powered Thermal Sources: A Case Study in Nut Processing

2.1 Dry roasting process modelling

Roasting process occurs in two different macro-phases. The first is the drying of the product, where most of the moisture content is evaporated at middle-low temperatures to stabilize the microstructure of the nut as reported by Young et al. [8] & Almonds board of California guidelines [4]; the latter is the roasting phase where the Maillard reaction takes place and gives the product the typical roasted and crunchy texture following Perren and Escher [9] & Saklar et al. works [10]. In this phase higher temperatures are used to enhance the roasted flavor while the moisture content is slightly decreased through diffusion.

Different kinds of nuts can be processed such as hazelnuts, almonds, peanuts, walnuts, and pistachios [11]. In this context, Almonds are considered as the main nut to be roasted as they are widespread in the dry roasting industry [1, 12-14] due to their widespread process conditions that are subjected to [1, 12-14] and their main properties have been assumed as in Table 1 [15].

Table 1. Main nut modelling assumptions

Following the manufacturer common process guidelines [5] the raw almonds enter in process with ambient temperature and moisture that ranges from 5% to 10% [4, 15, 16]. A moisture content of 8% is assumed from here below. From manufacturer process energy analysis, it is then assumed to have two different coefficients:

The former is for determining the heat loss of the main thermal source (e.g. the burner) and it is assumed at 90%. These losses are mainly due to the fact that not the whole thermal power by combustion that occurs in the burner is completely transferred and adiabatically mixed with the process air. A portion of the thermal power is lost due to convection and conduction losses through the main process duct [17].

The latter is for determing the heat exchange between the thermal medium and final product and it is assumed at 70%. This is due to the fact that, since the process air is flowing through the bed of product placed onto the main conveyor belt, not all the thermal energy of the air is transferred and so the exhaust air after the process is still warm. These losses can be reduced by optimizing the air flow path through the process camera but from manufacturer approach the thermal losses are around 30% resulting in a heat exchange/roasting coefficient of 70 % approximately as explained in Table 2.

Table 2. Main roasting process assumptions

In such roaster, three different process phases occur which are namely Drying, Roasting and Cooling. As shown in Figure 2, a temperature trend made with 3 different temperature dataloggers over the whole process in three different positions (left, right and central) onto the product conveyor belt is reported. As it is possible to notice the drying phase is a phase of transition where both evaporation of internal moisture and increase of surface kernel temperature occur. During the roasting phase the temperature is then stabilized at desired values to get the required aroma. Cooling phases at the end eventually needs to be done in order to stop the roasting reaction onto the dried nut and so to control the organoleptic features of the product. Indeed, if the product is not cooled enough the chemical cooking reactions go on and can ruin the desoldered characteristics of the process [4, 10].

qq20251117-164354.png

Figure 2. Temperature diagram example (made with datalogger in VUORMAR JUMBO process roaster) [5]

Following Figure 2 and Table 2 assumptions it is then possible to assess the overall required thermal power for the combined process. The evaluation can be made for each process phase to assess each thermal energy requirements according to assumptions made above. In the Table 3 below the whole roasting process is assessed.

Table 3. Main roasting process subdivision

The total necessary power is then given by the sum of latent heat and sensible heat as for the Eq. (1) computed for each phase (respectively drying and roasting) increase by means of the two effectiveness coefficient regarding the process explained above in Table 2. The cooling phase is not taken into account for thermal power computation, but it is shown to show the whole cycle dynamic.

$P_{ {total }}=\frac{P_{{latent }}+P_{{sensible }}}{\eta_{ {burner }} \cdot \eta_{{air product\ heat\ exchange }}}$                  (1)

Eq. (2) and Eq. (3) represent the latent and sensible heat power equations to be used above. The former is used for the evaporation of moisture content of the product given by the delta of $\left(M C_{{out }}-M C_{{in }}\right)$ (i.e. the relative humidity difference of the product at inlet and outlet of the process) considering a $h_{{vaporization }}$ latent heat of vaporization of water of 2272 kJ/kg. The latter is used for computing the sensible heat to be applied to the net product (whose specific heat is given by $cp_{ {product}}$) to increase its temperature from the inlet one ($T_{i n}$) to the outlet one ($T_{{out }}$) at the end of each phase following what written in Table 3.

$P_{ {latent }}=\dot{m}_{{product }} \cdot\left(M C_{{out }}-M C_{{in }}\right) \cdot h_{{vaporization }}$                  (2)

$P_{ {sensible }}=\dot{m}_{ {product }} \cdot c p_{{product }} \cdot\left(T_{{out }}-T_{ {in }}\right)$                    (3)

It is then possible to assess the specific power requirements referred to the unit of mass of product entering in the process by dividing the whole thermal power computed by Eq. (1) by the mass flow rate of product to be processes. As below, the specific power is function of the following parameters: Moisture content variation, Temperature increase and specific heat of the product (as the water latent heat of vaporization is a constant) as in Eq. (4).

$P_{ {specific }}=\frac{P_{ {total }}}{\dot{m}_{{product }}}$                        (4)

2.2 Hybrid gas – electric installation

Coming to the hybrid system features the electric heater acts as a pre-heating stage thus reducing the enthalpy leap demanded by the gas burner. Usually, common electric heaters are made by electric filaments with variable diameter and length as in the figure below. The heater is then installed as a common duct into the main process air duct.

The thermoregulation dynamic of the two systems presented is very different: while the gas burner is very rapid and clear in the transitory phases due to an analogic control in 4-20 mA of the main gas flow valve, the electric pre-heater follows a more discrete thermoregulation curve, that eventually can shut of only some heaters stages when modulating the power by means of solid state relays in the main electrical control cabinet.  In Figure 3 a simplified process flow diagram of the hybrid installation is shown.

qq20251117-164611.png

Figure 3. PFD (simplified process flow diagram) of hybrid dry roasting

Depending on the actual size of the heater, it is possible to assess the enthalpy leap of the process air entering the roaster respectively in Eq. (5) and Eq. (6). Namely the ration between these two values is given by the temperature increase to the process air.

$P_{ {pre-heating-stage }}=\dot{m}_{ {process \ air }} \cdot d h_{{electric \  heater }}$                      (5)

$P_{{burner-stage }}=\dot{m}_{ {process \  air }} \cdot d h_{ {burner }}$                     (6)

Coming to the assessment of the size of the electric pre-heater, it is then possible to use Eq. (7) to evaluate the actual gas savings in terms of standard cubic meters (smc) consumption over the electric installed pre-heater power.  Basically, it has been assessed how much gas flow volumetric rate can be saved by replacing it as main thermal source with the daily electrical energy producibility. The value is then normalized by dividing it by the installed pre-heater size to have an insight into gas saving (or replacing) efficiency. A higher value means that the available electric power is well exploited to save as much as possible while a lower one means that not the whole electric potential is being exploited enough.

$\begin{aligned} & { Saving }_{ {specific }} =\frac{{ smc }_{{avoidedgas }} \cdot{ hour }_{{work }} \cdot { days }_{ {work }}}{P_{ {pre-heater }}}\end{aligned}$                    (7)

While the absolute saving amount in €/year of avoided supplied gas can be assessed by means of Eq. (8) where the daily amount of saved gas can be recollected to the overall revenues:

$\begin{aligned} { Saving }_{{absolute }}= { smc }_{ {avoidedgas }} \cdot { hour }_{ {work }} \cdot { days }_{ {work }} \cdot{ Gas }_{{price }}\end{aligned}$                       (8)

Moreover from this daily approach it is possible to assess two other fundamental coefficients for the assessment: The former is the electric energy replacing factor of the process described by the Eq. (9). This is a coefficient that basically assess how much gas powered thermal source can be replaced by the electric energy use based on the fact that a portion of gas consumption is replaced by available electric energy coming from the PV in relation to a reference production output (which is for the considered JUMBO roaster 2000 kg/h [5]). A 100% replacing factor means that all the previous gas thermal source can be replaced by the use of available energy.

$Replacing\ Factor { }_{ {gas } \rightarrow { electric }}=\frac{S m c_{ {avoidedgas }}}{L H V_{N G} \cdot P_{{total }}}$                        (9)

Instead, the latter is the percentage exploitation of the heater in relation to the actual energy production availability which is represented by Eq. (10):

$Pre - Heater _{\% { Use }}=\frac{\left(\frac{E_{ {daily }}}{{ hour }{ }_{ {work }}}\right)}{P_{{pre }- { heater }}}$                   (10)

Table 4 sums up the considerations that have been made regarding the above-mentioned assumptions:

Table 4. Main general assumptions

2.3 Photovoltaic energy producibility

The overall producibility of Photovoltaic has been determined by means of the medium irradiation throughout the 2020-2024 period considered. In Figure 4 the medium irradiation for the North Italy Sample Town has been investigated by means of PV-GIS and local irradiation station database [6, 7].

qq20251117-165655.png

Figure 4. Medium irradiation 2020-2024

In particular, the following assumptions referred to in Table 5 have been made over the performance of the PV field: It has been considered a Panasonic Anchor 540 W power module, with a net electric efficiency referred to STC conditions of 21.2% and a Module Ventilation Efficiency of 70% (which is according to the study [20] the worst and precautionary ventilation coefficient). Then to assess the overall area used for the PHV field it has been considered to use a 90% coefficient:

The daily energy producibility has been evaluated by means of the Eq. (11) which is referred to the UNI/TS 11300 [20],

$E_{{daily }}=\frac{I_{{day }} \cdot P_{{photovoltaic }} \cdot M V E}{{ I\_reference }}$                    (11)

Table 5. Main PHV efficiency assumptions [21]

As it is clearly understandable, the photovoltaic gives its maximum contribution to the production during the spring and summer months such as from April to September. During winter and fall, the actual production difference is less evident.

The hourly production has been then evaluated through samples months such as January, April, July and October 2024 in all three-power size to get an insight of the possible system dynamics throughout the daily production cycle. Figure 5 below shows the 3 MW case scenario.

qq20251117-165854.png

Figure 5. Hourly production rate 3 MW

It has been then considered dedicating 100% of the produced energy to feed the electric pre-heater throughout the daily production cycle. It was then possible to compare the actual kWh energy daily production with the overall process specific energy requirements to recover the final maximum daily productivity of the roaster following Eq. (12).

$\dot{m}_{{hour }- { maximum }}=\frac{E_{{daily }}}{P_{{specific }}}$                         (12)

The aim of the computations above has been to investigate the feasibility of disposing of a hybrid thermal solution to generate the hot air used for dry roasting processes whenever a photovoltaic backup field is available. The size of the electric pre-heater follows the actual size of the photovoltaic field and can determine whether the heater is well exploited or not during actual use throughout the whole production year. The actual size follows indeed the specific energy consumption per kg of net roasted product that is ultimately determined by the required process parameters according to the product recipes. This can be also determined by the flexibility requirement of the customer who can decide to increase or decrease the power size according to variations in the production output and use of the roaster throughout the year to absorb some production output peak requests.

The economic benefit of such a system follows the country energy trading scenario and can be considered favorable whenever the electricity price is lower than the gas purchase cost: in this case the on-site use of produced energy would be much more valuable than actual re-selling onto the grid.

Following Italy energy trading scenario indeed the exploitation of produced electric energy has not been virtually favorable since 2018 where the gas price was higher than electricity trade value and so resulting that in such background the most efficient use of such PHV produced energy could have been to be sold onto the grid.

Despite the economic benefit is variable and depending on country energy market trade values, the environmental benefit is positive considering that potentially such a system could lead to a considerable emission reduction that peaks at 140-ton CO₂/year for 3 MW and 58-ton CO₂/year for 1 MW scenario.

The proposed solution may be particularly game changing whenever the energetic scenario of the country (where the customer is sited) is favorable for electric exploiting rather than gas use in order to increase the energetic efficiency of the process behind the manufacturing of dry nut industry [24]. France for example would be a country scenario where the exploitation of such device could help reduce the overall CO₂ emissions due to gas burning processes even more since the country energy system is less reliant on gas sources that eventually lead to a decrease in electricity specific cost and an increase in its use in industrial processes [25, 26].

Eventually the proposed solution can be integrated as well into different roasting methods as reported by Bagheri et al. [27] to simultaneously improve the organoleptic composition of the nut and still reduce the CO₂ emissions by keeping a high system flexibility.

The presented scenario could be eventually improved by modelling all the other utilities installed onto the case study factory and their management to optimize the hybrid feeding requirements. Moreover, a detailed model of the thermodynamic hybrid system can be delineated to optimize the burner stage and electric pre-heater sizes according to their interaction dynamic.