Design of a Technical Decision-Making Strategy to Collect Biomass Waste from the Palm Oil Industry as a Renewable Energy Source: Case Study in Colombia

Abstract

This work presents an effective design of a strategy to manage biomass waste (empty fruit bunch—EFB, kernel shell, and fiber) available from the processing of oil palm (Elaeis guineensis) in Colombia as a renewable energy source. This type of study is conducted for the first time in the country, and the proposed strategy is structured in four phases. Firstly, an inventory of available biomass waste was prepared based on information from 45 African palm oil companies of the approximately 70 that exist in the country. It was determined that the country had about 2762 kt of available waste (63.64% EFB, 12.55% kernel shell, and 23.81% fiber) for the year 2023. The estimates were conducted using a model that correlates processing capacity, the biomass generated, and the biomass demanded. The validation was performed using national reports. Subsequently, the minimum number (six) of storage centers in Colombia, where the largest amount of biomass can be stored, was determined. The center of gravity method was used to find the geographical location of each bulk storage center (municipality of Aracataca, Agustín Codazzi, San Martín, Puerto Wilches, Castilla La Nueva, and Cabuyaru). The next step was to determine the transportation costs as a decision criterion to select the best bulk storage center. When the required storage capacity does not exceed 211 kt·year⁻¹, Agustín Codazzi is the best option because it has the lowest transportation cost (USD 1.01·t⁻¹). When the storage capacity requirements exceed 211 kt·year⁻¹ but are less than 423 kt·year⁻¹, then Puerto Wilches and/or Aracataca are the best options (transportation cost of USD 1.7·t⁻¹). In all cases, Cabuyaru has the highest costs (USD 6.56·t⁻¹). Finally, an energy potential of 50,196 × 10⁶ GJ·year⁻¹ for the collected biomass was estimated, which makes this kind of waste an environmental alternative that could replace coal in Colombia.

1. Introduction

Colombia’s energy transition policies are geared toward clean energy. In this regard, agriculture plays a significant role because biomass waste has an energy potential that can be harnessed thermally, electrically, or as fuels [1]. This becomes even more important when considering the fact that the generation of agro-industrial waste in various stages of production processes is often a problem for industry, as these are not always adequately disposed of, leading to potential environmental impact. Therefore, designing strategies to give value to agro-industrial waste is crucial to promote circular economy schemes and sustainable development in a country or region [2].

In Colombia, inventories have been compiled to quantify the amount of biomass waste generated in various agricultural chains [1,3,4,5,6,7,8]. For the first semester of 2019, it was estimated that out of the 2,020,662 Agricultural Production Units (UPAs) reported by the National Administrative Department of Statistics [9], only 5.6% (13,157 UPAs) reported the utilization of their waste in agro-industrial activities; out of this quantity, 83% used their waste for composting, 15% for animal feed, and 2% for electricity, fuel, or heat generation [9]. Despite the low percentage of use for energy, it has been possible to estimate how much electricity can be generated by harnessing the available biomass in the country (15 GW for 2013) [6,8]. However, there are technical, economic, government, policy, and social barriers that make biomass utilization financially unviable [7,8,10,11,12,13,14]. One major barrier in Colombia is that a significant portion of agricultural biomass waste is produced in remote areas, directly impacting transportation costs [8]. Hence, designing biomass storage strategies to enable subsequent use is a critical aspect to effectively manage the final disposal of waste [10].

Many studies on biomass supply chains for energy purposes have been reported in the literature [15,16,17,18,19,20]. Some are based on multi-objective optimization models that aim to minimize costs while maximizing social and environmental benefits [21,22,23,24,25,26]; others include additional criteria. For instance, Zaharee et al. [27] integrate intermodal transportation (trucks and trains) into the biomass supply model according to biomass demand, a strategy that helps reduce gaseous pollutant emissions from transportation. Costa et al. [28] assess the technical, environmental, and economic impact of biomass logistics in the context of a hybrid energy system (heat and electricity). Gital-Durmaz and Bilgen [29] suggest that, in designing biomass supply chains, both the transportation distance and the available quantity of waste should be considered. Additionally, the authors recommend that the design of the supply chain should follow a strategy that includes integrating the geographic locations of biomass sources, employing a hierarchical approach with a multi-objective linear programming model aimed at maximizing profitability and minimizing the distance between collection points and power generation facilities.

Stochastic models using genetic algorithms have been employed for random decision making in biomass supply [30]. While various models exist, most studies treat biomass supply and demand as uncertain parameters. Additionally, other influential factors—such as political regulations, government subsidies, biomass market dynamics, and oil prices—are often difficult to quantify and therefore frequently overlooked [20]. No method is better than the other; the choice depends on the need and the information available to create the biomass storage strategy. This means that not all methods are used simultaneously; the method chosen will always depend on the scenario. Therefore, logistics strictly depend on factors such as location, costs, and social or environmental aspects.

In the case of palm oil production plants, the combustion of residual biomass is a feasible solution for local thermal energy needs when demanded. However, surplus biomass waste often exceeds on-site requirements, leading to untapped energy and material valorization potential. To optimize its utilization, it is crucial to design straightforward and effective methodologies to identify strategic collection points for these excess residues; transport and logistics could account for up to 90% of the total feedstock cost [31].

Integrating biomass waste into broader energy systems—such as power generation, polygeneration, or biorefinery schemes—can significantly enhance economic returns and environmental sustainability. These approaches support a circular economy by enabling the production of biofuels, biochemicals, and other high-value-added products, maximizing the efficiency of biomass resource management [32].

To achieve this, it is essential to assess the optimal conversion pathway for transforming biomass into the desired product. In this regard, thermochemical processes—primarily gasification and pyrolysis—offer a diverse range of valuable outputs. Moreover, hybrid scenarios that integrate thermochemical and biochemical conversion routes can further improve the efficiency of biomass waste utilization [32]. Regardless of the chosen optimal solution, the key bottleneck remains the supply chain and logistics required to transport biomass to the processing site.

In this study, a biomass supply strategy is proposed to establish a central biomass storage district based on criteria to group the largest amount of biomass within a smaller radius of action and generate the lowest transportation cost. The case study considers biomass waste (EFB, fiber, and kernel shell) generated from the industrial processing of oil palm fresh fruit bunches (FFBs) in Colombia. This type of work has not been carried out in the country and sets a roadmap for the utilization of biomass waste in Colombia, becoming a fundamental tool for decision makers in the agro-industrial sector of the country and for shaping public policy regarding the utilization of this type of waste.

2. Materials and Methods

To compile an inventory of the available waste in oil palm companies in Colombia, 45 out of approximately 70 companies in the country were selected. The selection criteria included choosing companies with the most significant impact on national palm oil production. To implement the proposed strategy, the algorithm presented in Figure 1 is suggested.

Step 1. Locating the available biomass quantity. It is necessary to determine the amount of available biomass waste, identify the locations where it is disposed of, and obtain Cartesian coordinates for these points. The oil palm industry in Colombia has the particularity of using some of its biomass waste for internal energy generation processes. Therefore, to ascertain the quantity of available waste, it is essential to understand how much biomass is consumed internally by the companies and how much remains available. For this reason, a mathematical model (Equations (1)–(7)) was developed. It incorporates variables (required steam flow and required energy flow) and constants (boiler efficiency and calorific value of the biomass used) to estimate the biomass consumed in an oil palm industry boiler and the remaining biomass. Additionally, in the model, it is necessary to consider the value of the parameters $k_1$ and $\eta$. The first parameter ($k_1,$ assumed to be 0.5) is a relationship that is handled experimentally in the industrial sector of oil palm processing, and it represents the tons of steam per ton of FFB. The second ($\eta$, assumed to be 0.85) represents the efficiency of the boiler, and to determine its value, the type of boiler normally used in African palm oil companies in Colombia was considered.

$${\dot{m}}_{R_V}=k_1{\dot{m}}_X$$

(1)

$${\dot{Q}}_{R_V}=k_2{\dot{m}}_{R_V}$$

(2)

$${\dot{Q}}_{E_N}={\displaystyle \frac{{\dot{Q}}_{R_V}}{\eta }}$$

(3)

$${\dot{m}}_{F_R}={\displaystyle \frac{k_3{\dot{Q}}_{E_N}}{100C_{P_F}}}$$

(4)

$${\dot{m}}_{C_R}={\displaystyle \frac{k_3{\dot{Q}}_{E_N}}{100C_{P_C}}}$$

(5)

$${\dot{m}}_{T_R}={\dot{m}}_{F_R}+{\dot{m}}_{C_R}$$

(6)

$${\dot{m}}_{T_D}={\dot{m}}_{T_G}-{\dot{m}}_{T_R}$$

(7)

where ${\dot{m}}_{R_V}$ is the required steam flow in t·h⁻¹; ${\dot{m}}_X$ is the amount of processed FFB per company in t·h⁻¹; ${\dot{Q}}_{R_V}$ is the energy flow in steam in kcal·h⁻¹; $k_2$ equals 0.64 Gcal ·t⁻¹ of steam; ${\dot{Q}}_{EN}$ is the required energy flow in kcal·h⁻¹; ${\dot{m}}_{F_R}$ is the required fiber consumption in t·h⁻¹; ${\dot{m}}_{C_R}$ is the required kernel shell consumption in t·h⁻¹; $C_{P_F}$ is the calorific value of the fiber, that is, 2.8 Gcal·t⁻¹ of fiber [33]; $C_{P_C}$ is the calorific value of the kernel shell, that is, 4.5 Gcal·t⁻¹ of kernel shell [34] (these data are on a dry basis, and the moisture percentages are ∼55%, ∼7%, and ∼32%, respectively [33,34]). $k_3$ is the supply percentage (60% for fiber and 40% for kernel shell); ${\dot{m}}_{T_R}$ is the total biomass required by the company for use in boilers in t·h⁻¹; ${\dot{m}}_{T_G}$ is the total biomass generated by each company in t·h⁻¹; and ${\dot{m}}_{T_D}$ is the total biomass available in each company (kernel shell, fiber, and EFB) in t·h⁻¹. It is important to keep in mind that all variables with a dot on top represent flows or are expressed per unit of time.

To solve the model, it is necessary to know the quantities of FFB processed by companies $\left({\dot{m}}_X\right)$. Therefore, an inventory of the amount of FFB processed was compiled in 45 companies out of the approximately 70 existing in the country. The biomass residue generated $\left({\dot{m}}_{T_G}\right)$ was estimated, considering that the percentage is around 45% (7% kernel shell, 14% fiber, and 24% EFB) [35,36,37,38]. From these percentages, the generated biomass residue was calculated, and with the model described above (Equations (3)–(7)), the biomass waste available per company was estimated $\left({\dot{m}}_{T_D}\right)$.

Step 2. Biomass grouping. After establishing the geographical location of the available biomass, groups that concentrate the highest amount of biomass within the smallest radius of action must be formed, and a central bulk storage point must be determined for each group. To determine the geographical location of a central bulk storage point, the center of gravity method is proposed. This method allows locating a central facility considering existing ones [31]. To achieve this, the quantities of biomass waste with their respective coordinates should be taken, and Equations (8) and (9) should be solved for each group.

$$X={\displaystyle \frac{\sum _{i=1}^n{X}_i{\dot{m}}_i}{\sum _{i=1}^n{\dot{m}}_i}}$$

(8)

$$Y={\displaystyle \frac{\sum _{i=1}^n{Y}_i{\dot{m}}_i}{\sum _{i=1}^n{\dot{m}}_i}}$$

(9)

where ${\dot{m}}_i$ is the quantity of waste (t·year⁻¹); $X_i$ and $Y_i$ are the Cartesian coordinates where the biomass is located; $X$ and $Y$ are the coordinates of the new storage center (main warehouse) or the potential company that will use the biomass.

Step 3. Final location decision. The final decision should consider factors such as location (road conditions, biomass dispersion, proximity to customers–suppliers, proximity to the main plant, ports, free trade zones, safeguarded or nature reserve zones), costs (biomass, transportation, productivity, labor, services, taxes, leasing, or land prices), and/or social aspects (availability of labor, partnerships, social security, environment, and climate).

Step 4. Usage alternatives. The criteria for choosing biomass to generate value-added products must be established. The utilization criteria can be biological (ethanol, hydrogen, sugars, composting, chemicals, CO₂ capture), chemical (cellulose, pellets, plastic additives, insulating materials, binders, paper), thermal (energy, activated carbon, hydrogen, pyrolytic oil, biochar), or combinations of the mentioned criteria (polygeneration system) [5,39,40,41,42].

This study proposes the use of oil palm biomass waste from an energy perspective (combustion). To do so, its theoretical energy potential was estimated using Equation (10). Additionally, to estimate the amount of carbon dioxide generated during biomass combustion (Equation (13)), it is necessary to establish the amount of carbon in the biomass in terms of mass (Equation (11)) and moles (Equation (12)). Finally, to obtain Equation (13), it is necessary to consider that, in a complete combustion reaction, the molar ratio of C to CO₂ is 1:1, as observed in the following reaction:

$$C+O_2\to C{O}_2$$

$${\dot{Q}}_i=1000{\dot{m}}_i{P}_{C_i}$$

(10)

$${\dot{m}}_{C_i}={\displaystyle \frac{ 1000{\dot{m}}_i{\%}_{C_{T_i}}}{100}}$$

(11)

$${\dot{mol}}_{C_i}={\displaystyle \frac{{\dot{m}}_{C_i}}{W_{M_C}}}$$

(12)

$${\dot{m}}_{{\left({CO}_2\right)}_i}={\dot{mol}}_{C_i}{W}_{M_{{CO}_2}}$$

(13)

where ${\dot{Q}}_i$ is the amount of energy supplied by biomass (kJ·year⁻¹); ${\dot{m}}_i$ is the amount of biomass (t·year⁻¹); 1000 is the conversion factor from kg to t; $P_{C_i}$ is the higher heating value of biomass (kJ·kg⁻¹); ${\dot{m}}_{C_i}$ is the amount of carbon in the studied biomass (kg·year⁻¹); ${\%}_{C_{T_i}}$ is the percentage of total carbon in the biomass; ${\dot{mol}}_{C_i}$ is the amount of carbon in the studied biomass (kmol·year⁻¹); $W_{M_C}$ is the molecular weight of carbon (12 kg·kmol⁻¹); ${\dot{m}}_{{\left({CO}_2\right)}_i}$ is the amount of carbon dioxide generated during the combustion of biomass (kg·year⁻¹); $W_{M_{{CO}_2}}$ is the molecular weight of carbon dioxide (44 kg·kmol⁻¹).

Comparing the energy potential of biomass waste (Equation (10)) with that of a fossil fuel, such as coal, Equation (14) can be used to calculate how much coal is required to supply the energy contained in the biomass. Finally, to calculate the amount of carbon dioxide that can be generated during the complete combustion of this coal (Equation (17)), just like in the biomass, it is necessary to establish the amount of carbon in coal in terms of mass (Equation (15)) or moles (Equation (16)).

$${\dot{m}}_{coal}={\displaystyle \frac{{\dot{Q}}_i }{P_{C_{coal}}}}$$

(14)

$${\dot{m}}_{C_{coal}}={\displaystyle \frac{{\dot{m}}_{coal}{\%}_{C_{T_{coal}}}}{100}}$$

(15)

$${\dot{mol}}_{C_{coal}}={\displaystyle \frac{{\dot{m}}_{C_{coal}}}{W_{M_C}}}$$

(16)

$${\dot{m}}_{{\left({CO}_2\right)}_{coal}}={\dot{mol}}_{C_{coal}}{W}_{M_{{CO}_2}}$$

(17)

where ${\dot{m}}_{coal}$ is the amount of coal needed to replace the energy contained in the biomass (kg·year⁻¹); $P_{C_{coal}}$ is the higher heating value of coal (kJ·kg⁻¹); ${\dot{m}}_{C_{coal}}$ is the amount of carbon in coal (kg·year⁻¹); ${\%}_{C_{T_{coal}}}$ is the percentage of total carbon in coal; ${\dot{mol}}_{C_{coal}}$ is the amount of carbon in coal (kmol·year⁻¹); and ${\dot{m}}_{{\left({CO}_2\right)}_{coal}}$ is the amount of carbon dioxide generated during coal combustion (kg·year⁻¹).

3. Results and Discussion

3.1. Location of Available Biomass Waste from the Oil Palm Industry

In the country, biomass waste (kernel shell, fiber, and EFB) from fresh fruit bunches (FFBs) is generated at 70 points across the national territory, corresponding to the locations of oil palm extraction companies. Figure 2 shows the map of Colombia, illustrating that the country’s palm oil industry is concentrated in six clusters (blue circles), each with a service radius of less than 400 linear kilometers. Some of the biomass waste (kernel shell and fiber) generated during the palm oil extraction process is used for energy purposes in the boilers of these companies. The remaining biomass is designated for commercialization or internal use within companies. Therefore, the amount of biomass used in boilers depends on their capacity and determines the available biomass.

The solution to the model (Equations (1)–(7)) for companies in clusters I, II, and III is presented in

Table 1. FFB waste generated and available in cluster I.

Companies	FFB $\left({\dot{\mathit{m}}}_{\mathit{X}}\right)$ in t/h	Generated Biomass $\left({\dot{\mathit{m}}}_{{\mathit{T}}_{\mathit{G}}}\right)$ in t/h				Required Steam Flow $\left({\dot{\mathit{m}}}_{{\mathit{R}}_{\mathit{V}}}\right)$ in t/h	Energy Flow Steam $\left({\dot{\mathit{Q}}}_{{\mathit{R}}_{\mathit{V}}}\right)$ in kcal·h−1	Required Energy Flow $\left({\dot{\mathit{Q}}}_{{\mathit{E}}_{\mathit{N}}}\right)$ in kcal·h−1	Required Biomass $\left({\dot{\mathit{m}}}_{{\mathit{T}}_{\mathit{R}}}\right)$ in t/h				Available Biomass $\left({\dot{\mathit{m}}}_{{\mathit{T}}_{\mathit{D}}}\right)$ in t/h
		EFB	Kernel Shell	Fiber	Subtotal				EFB	Kernel Shell	Fiber	Subtotal	EFB	Kernel Shell	Fiber	Subtotal
A	37.09	8.55	2.85	5.29	16.69	19	11,868,293	13,962,697	-	1.16	2.09	3.26	8.55	1.69	3.20	13.43
B	38.20	8.81	2.94	5.45	17.19	19	12,224,975	14,382,323	-	1.20	2.16	3.36	8.81	1.74	3.29	13.84
C	35.36	8.15	2.72	5.04	15.91	18	11,314,338	13,310,985	-	1.11	2.00	3.11	8.15	1.61	3.05	12.80
D	19.49	4.49	1.50	2.78	8.77	10	6,236,075	7,336,559	-	0.61	1.10	1.71	4.49	0.89	1.68	7.06
E	53.42	12.31	4.10	7.62	24.04	27	17,093,335	20,109,806	-	1.68	3.02	4.69	12.31	2.43	4.61	19.35
F	33.96	7.83	2.61	4.85	15.28	17	10,867,408	12,785,186	-	1.07	1.92	2.98	7.83	1.54	2.93	12.30
G	29.68	6.84	2.28	4.23	13.36	15	9,496,956	11,172,890	-	0.93	1.68	2.61	6.84	1.35	2.56	10.75
H	9.67	2.23	0.74	1.38	4.35	5	3,095,574	3,641,851	-	0.30	0.55	0.85	2.23	0.44	0.83	3.50
I	5.55	1.28	0.43	0.79	2.50	3	1,774,754	2,087,946	-	0.17	0.31	0.49	1.28	0.25	0.48	2.01
J	13.44	3.10	1.03	1.92	6.05	7	4,300,436	5,059,337	-	0.42	0.76	1.18	3.10	0.61	1.16	4.87
K	69.87	16.11	5.37	9.97	31.44	35	22,359,905	26,305,771	-	2.19	3.95	6.14	16.11	3.18	6.02	25.31
L	4.65	1.07	0.36	0.66	2.09	2	1,487,398	1,749,881	-	0.15	0.26	0.41	1.07	0.21	0.40	1.68
M	32.58	7.51	2.50	4.65	14.66	16	10,424,702	12,264,356	-	1.02	1.84	2.86	7.51	1.48	2.81	11.80
N	9.72	2.24	0.75	1.39	4.38	5	3,111,267	3,660,314	-	0.31	0.55	0.85	2.24	0.44	0.84	3.52
O	15.33	3.53	1.18	2.19	6.90	8	4,904,517	5,770,020	-	0.48	0.87	1.35	3.53	0.70	1.32	5.55
Total	408	94.04	31.35	58.21	183.60	204	130,559,935	153,599,923	-	12.80	23.04	35.84	94.04	18.55	35.27	147.76

,

Table 2. FFB waste generated and available in cluster II.

Companies	FFB $\left({\dot{\mathit{m}}}_{\mathit{X}}\right)$ in t/h	Generated Biomass $\left({\dot{\mathit{m}}}_{{\mathit{T}}_{\mathit{G}}}\right)$ in t/h				Required Steam Flow $\left({\dot{\mathit{m}}}_{{\mathit{R}}_{\mathit{V}}}\right)$ in t/h	Energy Flow Steam $\left({\dot{\mathit{Q}}}_{{\mathit{R}}_{\mathit{V}}}\right)$ in kcal·h−1	Required Energy Flow $\left({\dot{\mathit{Q}}}_{{\mathit{E}}_{\mathit{N}}}\right)$ in kcal·h−1	Required Biomass $\left({\dot{\mathit{m}}}_{{\mathit{T}}_{\mathit{R}}}\right)$ in t/h				Available Biomass $\left({\dot{\mathit{m}}}_{{\mathit{T}}_{\mathit{D}}}\right)$ in t/h
		EFB	Kernel Shell	Fiber	Subtotal				EFB	Kernel Shell	Fiber	Subtotal	EFB	Kernel Shell	Fiber	Subtotal
A	43.33	9.99	3.33	6.18	19.50	22	13,866,545	16,313,582	-	1.36	2.45	3.81	9.99	1.97	3.74	15.69
B	45.84	10.57	3.52	6.54	20.63	23	14,669,674	17,258,440	-	1.44	2.59	4.03	10.57	2.08	3.95	16.60
C	32.63	7.52	2.51	4.66	14.68	16	10,440,561	12,283,013	-	1.02	1.84	2.87	7.52	1.48	2.81	11.82
D	28.89	6.66	2.22	4.12	13.00	14	9,245,566	10,877,137	-	0.91	1.63	2.54	6.66	1.31	2.49	10.46
E	90.87	20.95	6.98	12.97	40.89	45	29,079,870	34,211,612	-	2.85	5.13	7.98	20.95	4.13	7.83	32.91
F	24.85	5.73	1.91	3.55	11.18	12	7,950,885	9,353,982	-	0.78	1.40	2.18	5.73	1.13	2.14	9.00
G	56.68	13.06	4.35	8.09	25.51	28	18,138,396	21,339,290	-	1.78	3.20	4.98	13.06	2.58	4.89	20.53
H	80.85	18.63	6.21	11.54	36.38	40	25,871,076	30,436,560	-	2.54	4.57	7.10	18.63	3.68	6.97	29.28
I	31.88	7.35	2.45	4.55	14.34	16	10,200,705	12,000,830	-	1.00	1.80	2.80	7.35	1.45	2.75	11.54
J	4.38	1.01	0.34	0.62	1.97	2	1,400,119	1,647,199	-	0.14	0.25	0.38	1.01	0.20	0.38	1.58
K	38.15	8.79	2.93	5.44	17.17	19	12,207,061	14,361,249	-	1.20	2.15	3.35	8.79	1.73	3.29	13.82
L	41.55	9.58	3.19	5.93	18.70	21	13,297,183	15,643,745	-	1.30	2.35	3.65	9.58	1.89	3.58	15.05
M	22.97	5.29	1.76	3.28	10.34	11	7,349,609	8,646,599	-	0.72	1.30	2.02	5.29	1.04	1.98	8.32
N	7.13	1.64	0.55	1.02	3.21	4	2,282,749	2,685,587	-	0.22	0.40	0.63	1.64	0.32	0.62	2.58
Total	550.00	126.77	42.26	78.48	247.5	275	176,000,000	207,058,824	-	17.25	31.06	48.31	126.77	25.00	47.42	199.19

and

Table 3. FFB waste generated and available in cluster III.

Companies	FFB $\left({\dot{\mathit{m}}}_{\mathit{X}}\right)$ in t/h	Generated Biomass $\left({\dot{\mathit{m}}}_{{\mathit{T}}_{\mathit{G}}}\right)$ in t/h				Required Steam Flow $\left({\dot{\mathit{m}}}_{{\mathit{R}}_{\mathit{V}}}\right)$ in t/h	Energy Flow Steam $\left({\dot{\mathit{Q}}}_{{\mathit{R}}_{\mathit{V}}}\right)$ in kcal·h−1	Required Energy Flow $\left({\dot{\mathit{Q}}}_{{\mathit{E}}_{\mathit{N}}}\right)$ in kcal·h−1	Required Biomass $\left({\dot{\mathit{m}}}_{{\mathit{T}}_{\mathit{R}}}\right)$ in t/h				Available Biomass $\left({\dot{\mathit{m}}}_{{\mathit{T}}_{\mathit{D}}}\right)$ in t/h
		EFB	Kernel Shell	Fiber	Subtotal				EFB	Kernel Shell	Fiber	Subtotal	EFB	Kernel Shell	Fiber	Subtotal
A	23.13	5.33	1.78	3.30	10.41	11.56	7,400,423	8,706,380	-	0.73	1.31	2.03	5.33	1.05	1.99	8.38
B	106.49	24.54	8.18	15.19	47.92	53.24	34,075,689	40,089,046	-	3.34	6.01	9.35	24.54	4.84	9.18	38.56
C	47.91	11.04	3.68	6.84	21.56	23.96	15,331,488	18,037,045	-	1.50	2.71	4.21	11.04	2.18	4.13	17.35
D	1.25	0.29	0.10	0.18	0.56	0.62	399,088	469,515	-	0.04	0.07	0.11	0.29	0.06	0.11	0.45
E	63.70	14.68	4.89	9.09	28.67	31.85	20,384,326	23,981,559	-	2.00	3.60	5.60	14.68	2.90	5.49	23.07
F	42.10	9.70	3.23	6.01	18.95	21.05	13,472,560	15,850,071	-	1.32	2.38	3.70	9.70	1.91	3.63	15.25
G	38.08	8.78	2.93	5.43	17.14	19.04	12,185,832	14,336,273	-	1.19	2.15	3.35	8.78	1.73	3.28	13.79
H	10.52	2.43	0.81	1.50	4.74	5.26	3,367,317	3,961,550	-	0.33	0.59	0.92	2.43	0.48	0.91	3.81
I	14.53	3.35	1.12	2.07	6.54	7.27	4,651,093	5,471,874	-	0.46	0.82	1.28	3.35	0.66	1.25	5.26
J	5.16	1.19	0.40	0.74	2.32	2.58	1,651,854	1,943,358	-	0.16	0.29	0.45	1.19	0.23	0.45	1.87
K	150.80	34.76	11.59	21.52	67.86	75.40	48,256,504	56,772,358	-	4.73	8.52	13.25	34.76	6.85	13.00	54.61
L	23.38	5.39	1.80	3.34	10.52	11.69	7,482,968	8,803,492	-	0.73	1.32	2.05	5.39	1.06	2.02	8.47
M	20.03	4.62	1.54	2.86	9.01	10.02	6,410,315	7,541,547	-	0.63	1.13	1.76	4.62	0.91	1.73	7.25
N	31.22	7.20	2.40	4.45	14.05	15.61	9,989,431	11,752,271	-	0.98	1.76	2.74	7.20	1.42	2.69	11.31
O	53.18	12.26	4.09	7.59	23.93	26.59	17,019,134	20,022,511	-	1.67	3.00	4.67	12.26	2.42	4.59	19.26
P	1.51	0.35	0.12	0.21	0.68	0.75	481,976	567,031	-	0.05	0.09	0.13	0.35	0.07	0.13	0.55
Total	633.00	145.90	48.63	90.32	284.85	316.50	202,560,000	238,305,882	-	19.86	35.75	55.60	145.90	28.77	54.57	229.25

, respectively. Cluster IV is not considered because of the low impact on national palm oil production. Clusters V and VI are not considered because waste is concentrated in only one company; they are isolated from the other clusters; and their quantity is small compared to those concentrated in clusters I, II, and III. The letters (A–O) in Table 1, Table 2 and Table 3 represent the names of the different companies. The data were collected through surveys conducted during visits to companies in the year 2019. Throughout the process, the confidentiality of the collected information was ensured. These surveys were designed in collaboration with field experts and tailored to the specific needs of our research, as this topic is of particular interest in shaping energy policy in Colombia, both at the public and private levels. With the waste contribution percentages from each of the companies and the latest official FFB production data reported to date in Colombia [43,44], an estimate of the waste available until 2023 was made (Figure 3). To achieve this, it was considered that the capacity of the companies has not grown significantly in the last ten years. It is worth noting that the estimation was not carried out beyond 2023 because, in Colombia, neither Fedepalma nor any other organization has reported more updated data yet.

To validate the estimates, only the values of kernel shell reported by Fedepalma were considered [43,44]. The country reports the national production of FFB as well as the data on available kernel shell, which is not the case for fiber and EFB. Figure 4 shows that the trends are well marked in clusters I and II; in cluster III, it is not as noticeable, which may be due to the percentages varying from company to company.

3.2. Biomass Waste Collection

As mentioned above, two criteria must be met for biomass collection: the first is grouping the largest amount of biomass over the shortest distance possible; the second is determining a central geographical collection point for each group. Therefore, to meet the first criterion, the Google Earth application was used to locate the companies (points where biomass is generated) and select those connected by roads at distances ranging from 90 to 160 km (

Table 4. Available biomass by groups considering the grouping of companies with distances between 90 and 160 km according to the estimates put forward for 2023.

Groups	Companies	Distance (km)	Available Biomass (kt·year−1)				Available Biomass (kt·year−1·km−1)
			EFB	Kernel Shell	Fiber	Subtotal	EFB	Kernel Shell	Fiber	Subtotal
1	10	120	269.44	53.14	100.78	423.37	2.25	0.44	0.84	3.53
2	4	90	134.41	26.51	50.28	211.2	1.49	0.29	0.56	2.35
3	4	129	232.32	45.82	86.9	365.04	1.8	0.36	0.67	2.83
4	9	108	313.02	61.73	117.08	491.83	2.9	0.57	1.08	4.55
5	10	160	445.31	87.82	166.57	699.7	2.78	0.55	1.04	4.37
6	6	115	363.7	71.73	136.04	571.47	3.16	0.62	1.18	4.97
Total			1758.21	346.75	657.65	2762.61	14.38	2.84	5.38	22.6

). Studies in Colombia on the collection of other biomass waste suggest circular distances of 150 km to consider the social and economic aspects of the study area [45], such as proximity to the customer, manpower availability, corporate alliances, security, transport costs, fixed costs, and productivity costs.

Table 4 shows the distribution of companies in groups according to their proximity, assuming that, on average, companies operate 4500 h per year. Group 6 concentrates the highest amount of biomass per kilometer (4.97 kt·year⁻¹·km⁻¹), and the number of companies is also lower (5) compared to group 5. The latter accumulates the highest amount of biomass per year (699.7 kt·year⁻¹), but it is dispersed over a greater distance (160 km). The data reported in Table 4 were calculated based on the biomass estimated for 2023 (Figure 3). It was considered that the national production [44] for that year was 406 t·h⁻¹, 530 t·h⁻¹, and 351 t·h⁻¹ (Figure 2) for clusters I, II, and III, respectively.

To meet the criterion to determine the central collection point for each group in Table 4, the center of gravity method was used. To achieve this, it was necessary to use the geographical coordinates (Cartesian mode) of the companies. Subsequently, Equations (8) and (9) were solved for each group, thus finding the coordinates of the central bulk storage points (Figure 5). The coordinates correspond to the municipalities of Aracataca—Magdalena (group 1), Agustín Codazzi—Cesar (group 2), San Martín—Cesar (group 3), Puerto Wilches—Santander (group 4), Castilla La Nueva—Meta (group 5), and Cabuyaro—Meta (group 6). Additionally,

Table 5. Distance from each of the companies to the collection center of each group.

Group 1		Group 2		Group 3		Group 4		Group 5		Group 6
Companies	Distance to Aracataca (km)	Companies	Distance to Codazzi (km)	Companies	Distance to San Martín (km)	Companies	Distance to P. Wilches (km)	Companies	Distance to Castilla (km)	Companies	Distance to Cabuyaru (km)
1	58.5	1	9	1	90	1	24	1	23.7	1	50.8
2	20	2	13.3	2	25.7	2	18.2	2	23.7	2	60
3	15	3	90	3	51.4	3	7.3	3	49	3	33.4
4	20	4	21.1	4	51.4	4	40.8	4	45	4	25.7
5	39	---	---	---	---	5	10.5	5	53.5	5	121
6	5	---	---	---	---	6	63	6	65.4	6	103
7	5	---	---	---	---	7	49.6	7	50.5	---	---
8	8	---	---	---	---	8	17.2	8	64.1	---	---
9	80	---	---	---	---	9	20.3	9	56.4	---	---
10	123	---	---	---	---	---	---	10	25.7	---	---

shows the connection distances between each company and its respective collection center.

With the groups and their central points, a window to the country is opened, showing that in Colombia, six key points can be considered for the operation of energy districts, polygeneration plants, or bulk biomass waste storage centers.

3.3. Final Decision

The cost factor was selected as the decision criterion, considering only the costs of raw materials (waste) and transportation. In the country, the unit costs of biomass per ton in 2023 were close to USD 5, USD 37.5, and USD 12.5 for EFB, kernel shell, and fiber, respectively. These costs and the amount of biomass per year (Table 4) were used to calculate the biomass costs (Table 5). To calculate transportation costs, factors such as the amount of biomass per year, the number of trips considering transportation in 25-ton trucks, loading time (2 h), unloading time (2 h), actual distance from each company to the central storage point, and the cost per kilometer of travel were considered. The latter varies for each group because it depends on variables such as tolls, travel hours, and road conditions. Many variables must be considered; therefore, the platform of the Ministry of Transportation of Colombia [46] was used to calculate the cost of transportation per kilometer in 2023.

Table 6. Estimated biomass and transportation costs.

Group	Central Point (Municipality)	Available Biomass (kt·Year−1)	Cost of Available Biomass (USD·Year−1)	Transport Cost
				USD·km−1	USD·Year−1	USD·t−1
1	Aracataca	423.36	4,599,730	1.25	711,429	1.68
2	Codazzi	211.20	2,294,620	1.19	212,363	1.01
3	San Martín	365.04	3,966,042	1.67	1,145,035	3.14
4	Puerto Wilches	491.83	5,343,610	1.35	839,166	1.71
5	Castilla La Nueva	699.70	7,602,006	2.33	3,474,633	4.97
6	Cabuyaro	571.47	6,208,861	2.06	3,748,222	6.56
Total		2762.6	30,014,869	9.85	10,130,852	19.05

shows that group 6 has the highest transportation cost per ton (USD 6.56·t⁻¹), but it does not accumulate the largest amount of biomass. However, group 2 has the lowest transportation cost per ton (USD 1.01·t⁻¹) but collects only 30% of the biomass of group 5 (maximum amount collected per group) and 37% of what group 6 collects.

Figure 6 illustrates the relationship between variable costs (associated with biomass and transportation) and biomass availability for each analyzed scenario. This graph serves as a decision-making tool, helping identify the most cost-effective location for a biomass utilization facility by considering raw material availability and scenario-specific associated costs. The selection of a facility’s location is fundamentally constrained by biomass availability within a given region.

The decision-making process for facility installation can follow two main approaches. The first is a product-driven approach, where a specific product and its required production capacity are defined, and the necessary biomass is then estimated to meet these targets. The second is a resource-driven approach, which identifies the available biomass and determines the most viable products and their corresponding scale. In this second approach, feasibility is strictly limited by the available biomass, and the data in Table 6 would only serve to compare the variable costs of utilizing the biomass in its local region against the costs in other analyzed regions.

In contrast, when the required biomass is predetermined based on a specific production target, Figure 6 suggests that for processing capacities of up to 211 kt·year⁻¹, establishing a collection center in Agustín Codazzi is the most cost-effective option. For capacities between 211 and 423 kt·year⁻¹, Puerto Wilches or Aracataca emerge as viable alternatives. Facilities processing between 423 and 500 kt·year⁻¹ would benefit from a collection center in Puerto Wilches, while for capacities exceeding 500 kt·year⁻¹, Castilla la Nueva offers the most economically favorable location.

It is essential to highlight that these recommendations are based solely on variable cost optimization. They do not account for fixed costs, potential economies of scale, or multi-site facility configurations, which could further refine investment decisions. Future studies should explore the feasibility of multiple collection centers and distributed processing facilities to enhance overall system efficiency.

3.4. Alternative Use

According to the data presented in Table 4, it is estimated that in 2023, approximately 2.76 × 10³ kt·year⁻¹ of biomass was available in the country, of which 63.64% was EFB, 12.55 % was kernel shell, and 23.81% was fiber. By relating these biomass quantities to their respective heating values (17,940 kJ·kg⁻¹ for EFB, 20,095 kJ·kg⁻¹ for kernel shell, and 17,770 kJ·kg⁻¹ for fiber [33,34]; all data are on a dry basis, and the moisture percentages are ∼55%, ∼7%, and ∼32%, respectively [33,34]), the energy potential of the available biomass is 31.54 × 10⁶ GJ·year⁻¹ for EFB, 6.97 × 10⁶ GJ·year⁻¹ for kernel shell, and 11.69 × 10⁶ GJ·year⁻¹ for fiber (Figure 7a). If we replace this energy with a non-renewable fuel like coal, around 1.56 × 10³ kt of coal would not be burned (Figure 7a). Additionally, comparing the combustion of available biomass and the mentioned amount of coal, 2.76 × 10³ kt·year⁻¹ of biomass would generate around 4.95 × 10³ kt·year⁻¹ of CO₂ (assuming that the percentage of total carbon in fiber, EFB, and kernel shell is 49.5%, 48.19%, and 51%, respectively), while 1.56 × 10³ kt of coal with 70% of total carbon would produce around 4.00 × 10³ kt·year⁻¹ of CO₂ (Figure 7b). As observed, the value is lower; however, unlike coal, in biomass combustion, the CO₂ cycle can be considered neutral because it is required by palm cultivation itself during growth; thus, it represents a reduction of around 80%·year⁻¹ in CO₂ (Figure 7c).

4. Conclusions

Biomass waste generated from palm oil processing in Colombia presents an opportunity to produce value-added products for energy purposes. In this context, it was determined that approximately 2762.6 kt of biomass was generated in 2023 by the country’s palm industries, with an energy potential of 50.20 × 10⁶ GJ, which remained available at the companies after the palm oil processing. Additionally, the study reveals the existence of six strategic points for bulk biomass storage in the country. Considering the capacities of the storage centers, the lowest biomass transportation cost is USD 1.01·t⁻¹, and it can be used to store up to 211 kt·year⁻¹ annually in the municipality of Codazzi. The highest transportation cost is USD 6.56·t⁻¹, suitable for storing up to 571.47 kt·year⁻¹ in the municipality of Cabuyaro. Finally, the thermal utilization of the available biomass is proposed as an alternative to replace coal and reduce the CO₂ emissions that would result from its combustion.

It must be mentioned that future studies should employ more detailed siting optimization algorithms to refine location selection. The center of gravity method was chosen due to its low data requirements, making it suitable for Colombia, where logistical cost data, infrastructure constraints, and demand variations remain incomplete. While more sophisticated optimization approaches require extensive datasets, the center of gravity method provides a practical and computationally efficient approximation, given the current data limitations. Additionally, this study does not quantify CO₂-equivalent emissions from transportation or incorporate a comprehensive life cycle analysis (LCA). Recognizing these aspects, it is important to emphasize the need for future research integrating efficiency assessments and environmental impact analyses. Despite these limitations, this paper constitutes one of the first references in the Colombian context, providing valuable baseline data for researchers, industry professionals, and policymakers.