Explorative Study of Aquaponics Systems in Indonesia

Abstract

Aquaponics is the combined cultivation of fish in aquaculture systems and plants in hydroponic systems. This study aimed to explore and evaluate new aquaponics system designs in Jakarta, Indonesia, in terms of water and nutrient efficiency, and/or profitability. In total, one brackish water system with barramundi and tomato (BW) and three freshwater systems with Nile tilapia, rice, and duckweed (FW + DW), Nile tilapia and rice (FW − DW), and Nile tilapia and tomato (FW-tomato) were investigated using dynamic mathematical models. The results showed an increase of N, P, and water use efficiency by 10%, 18%, and 31%, respectively, in FW-tomato compare to FW − DW. Meanwhile, FW + DW only slightly increased the efficiency but worsened the already negative net present value (NPV) of FW − DW due to the high land prices. The BW system had similar water and nutrient efficiency to FW-tomato, but lower NPV. Profit of the BW and FW-tomato becomes comparable if the hydroponic area in the BW system can be expanded by 35%, or the selling price of tomato or fish in the BW can be increased by 20%. Given the expected growth of both the production of tomatoes and aquaculture in Indonesia, there is potential for further introduction of aquaponics systems in Indonesia.

1. Introduction

Indonesia, the fourth most populous country globally, has food and nutrition security as a central issue. The Economist Intelligence Unit ranked Indonesia as the 72nd of 109 countries in the Global Food Security Index, which measures food safety, availability, quality, and affordability [1]. Indonesia is a net importer of grains, horticulture products, and livestock [2] even though Indonesia is known as an agrarian country in which almost 32% of the land is used for agricultural production and around 30% of the population is working in agriculture, forest, and fishery [3,4]. Moreover, the country is one of the biggest rice importers in the world [5], as rice is the staple food for most of the population in Indonesia. At the same time, the production of tomatoes in Indonesia increased from 12,000 tons in 1969 to 976,790 tons in 2018, growing at an average annual rate of 12.5% [6], and in 2019 it still showed a growth of 4.5% [7]. Moreover, it is expected that the aquaculture market will register a compound annual growth rate of almost 18% by 2023 [8].

Several issues in terms of agricultural practices in Indonesia negatively affect the food security index. Firstly, chemical and fertilizer usage in Indonesia is expanding [9]. Secondly, the land is overused for agriculture. It was recorded that Indonesia is one of the five leading countries of forest loss in percentages over the past decades, with the expansion of agriculture as one of the contributing factors [10]. Thirdly, most of the food is produced outside the city, while almost half of the Indonesian population lives in urban areas [11]. For instance, Jakarta can only provide less than 0.5% of rice demand locally [12]. Considering Indonesia as archipelago country, local food production will lower energy and cost spent for food distribution. Lastly, the overexploitation of groundwater is a worrying issue [2].

Given these issues, a shift to a more innovative and sustainable way of farming in Indonesia is vital. Aside from the sustainability aspect, the ability to produce food locally in an archipelago country like Indonesia is essential to have short supply chains and low production costs. Promoting sustainable agroindustry is one of the Indonesian government’s targets in the Grand Strategy of Agriculture Development 2015–2045 [13]. One alternative for sustainable food production is to modify the growth medium such that it can prevent water and land from fast depletion.

Currently, hydroponics, a soil-less type of greenhouse horticulture, is increasing in popularity, because of the efficient management of resources and food production. A wide variety of plant species can be grown in hydroponic systems, including leafy vegetables, fruits, tuber plants, and grain. This type of food production can save a large amount of water, produces a higher yield than conventional horticulture, and is hardly influenced by climate change [14]. Additionally, small to medium-sized hydroponic systems (HPS) can potentially be installed in the city to fulfil local food demand [15].

Having a sustainable fertilizer source is essential in HPS, since nutrient solutions are the primary inputs in HPS. An option to increase the nutrient use efficiency of HPS is by connecting a recirculating aquaculture system (RAS) to the HPS. Such food production technology is known as aquaponics, in which discharged water from RAS is managed such that it meets the plant’s nutrient requirement in the HPS. Conventional aquaponics uses a one-loop system, where partly demineralized water from HPS is sent back to RAS [16]. On one hand, this system can produce both fish and plants in an environmentally friendly way, allowing a straightforward installation process. On the other hand, this system is suboptimal because it equates the water characteristics in both HPS and RAS, while each plant and fish species has its own optimal water quality conditions [17].

In the current study, the first adjustment we propose is to transform the one-loop system into a decoupled aquaponics system (DAPS), recently also called an on-demand coupled system [18]. In DAPS, water is only transported from the RAS to HPS, so that both facilities can be optimized separately [19]. However, this adjustment increases the amount of fresh water required in RAS. Sustainability of DAPS can be increased by incorporating a digester in the system to produce energy and bio-fertilizer from solid waste material from the RAS [20]. The second modification we propose is to substitute fish feed partially by duckweed. The potential of duckweed utilization as an alternative to fish feed has been proven in many studies [21,22,23]. The main reason for including duckweed in the feed is because feed cost is one of the most pressing issues in aquaculture practices, including Indonesia. Feed covers more than 50% of the total production costs and the composition can reach up to 50% of proteins [24,25]. Typically, protein sources of fish feed come from fish meal, usually not locally available and expensive [24]. The second reason for inclusion of a duckweed cultivation system (DCS) is that duckweed cultivation can be a polishing step of the water discharged from the HPS. However, so far, there has been little discussion about the sustainability of aquaponics systems integrated with duckweed cultivation [26].

Limitations in practical and scientific knowledge make it challenging to design a close to optimal aquaponics system in Indonesia. The objective of the study was to explore and investigate aquaponics system designs in Jakarta, a city with more than 9 million inhabitants, such that high water, nutrients, and energy use efficiencies are obtained while still being economically feasible. The virtual aquaponics system, called the IndonesiAP system in what follows, uses low technologies to produce sustainable food. Such a choice is very relevant in a developing country because it can be easily adapted and requires low investment costs.

The first proposed design of an aquaponics system uses freshwater and produces Nile tilapia in the RAS and uses a rice paddy in the HPS. This arrangement is, in what follows, called freshwater (FW) system. Producing rice in aquaponics facilities that can be implemented in an urban area can be a viable option for sustainable food supply in Indonesia. Nevertheless, there is not much known about the performance of aquaponics systems that grow rice in an HPS facility. In 2017, a farmer community in East Java, called Bengkel Mimpi, constructed a small scale one-loop aquaculture system integrated with a hydroponic rice paddy. Their design became an inspiration for the designs of the IndonesiAP systems.

Since IndonesiAP is virtually located in Jakarta, which is near a coastline, this study also explored and evaluated an aquaponics system design under brackish water (BW) conditions (total dissolved solid of brackish water is between 1000 to 10 000 ppm [27]. Asian barramundi (Lates calcarifer) and tomato were chosen to be evaluated in the BW IndonesiAP system. In total, four systems were studied, namely three freshwater systems that are: FW + DW (Nile tilapia, rice, and duckweed), FW − DW (Nile tilapia, rice without duckweed), FW tomato (Nile tilapia and tomato), and one brackish water system BW (barramundi and tomato).

This study aimed to explore possibility of AP system in Indonesia. The purpose of this research study is to gain insight into the dynamics and interactions of the IndonesiAP system. It also develops an understanding of the system’s sensitivity, which is essential for nutrient, water, energy, and economy optimization. In practice, the objective of this study is to create a model of nutrient, water, and energy balance of IndonesiAP system and using it to evaluate the sustainability performance and perform sensitivity analysis of the system. Finally, it is used to find optimal system design parameters.

2. Materials and Methods

2.1. System Description

IndonesiAP systems, located in Jakarta (Indonesia), consist of a combination of RAS, HPS, duckweed cultivation system (DCS), and anaerobic digester unit (ADU). The description and assumptions used for the ADU follow from Dijkgraaf et al. (2019) [19]. An overview of the system is shown in Figure 1. In FW − DW systems, duckweed production ditches and pelletizer unit are disregarded.

2.2. Recirculating Aquaculture Aystem

For the fish production, the RAS configuration, sizing, and operation is similar as proposed by Dijkgraaf et al. (2019) [20]. The RAS consists of five different elements, namely four equal fish tanks with total volume of 12.5 m³, a settling tank, a sump, a moving bed biofilm reactor, and a trickling filter. Fish production is staggered and the fish is harvested every 200 days. The maximum stocking density in the fish tanks is 40 kg/m³.

Groundwater is the main water source of the system. Characteristics of the water in the fish tanks should meet the ranges shown in

Table 1. Water requirements for Nile tilapia and barramundi.

Requirement	Unit	Nile Tilapia		Barramundi
		Value	Source	Value	Source
Temperature	°C	27–29	[28]	26–32	[29]
pH		6–9	[28]	7.5–8.5	[29]
Ammonia (NH3-N)	mg·L−1	<1	[28]	<0.46	[30]
Nitrite (NO2-N)	mg·L−1	<5	[28]	<14.5	[30]
Nitrate (NO3-N)	mg·L−1	<300–400	[28]	<150	[31]
Salinity	µS/m			0–22.4	[30]

to achieve successful tilapia and barramundi production. Jakarta groundwater characteristics are presented in

Table 2. Characteristics of Jakarta groundwater.

Parameter	Unit	Value	Source
Temperature	°C	29–29.5	[32]
pH		6.5–7	[32]
Ammonium (NH4+)	mg·L−1	2.8	[33]
Ammonia (NH3-N)	mg·L−1	8.6	[34]
Nitrite (NO2-N) and nitrate (NO3-N)	mg·L−1	0.29	[33]
Salinity (west part of West Jakarta)	dS/m	2.8	[35]

. As groundwater temperature (29–29.5 °C) is in the acceptable range for tilapia and barramundi, heating or cooling groundwater is unnecessary.

Calculating the feed input is essential as it determines the nutrient dynamics of the system. The feed requirement is determined on the basis of the fish growth model of Timmons and Ebeling (2010) [36]. In the model, fish growth parameters for tilapia follow from Dijkgraaf et al. (2019) [20], while for barramundi, the growth parameters are presented in Appendix A, Table A1.

The model of Timmons and Ebeling (2010) [36] predicts a daily increase of fish length, which is then used to estimate the weight of the fish. The weight of the fish determines subsequently the amount of feed required, which is determined by the feed conversion ratio (FCR). The sigmoid function, as used by Dijkgraaf et al. (2019) [20], was used to calculate the FCR of tilapia. Meanwhile, the FCR for barramundi was defined as:

$$\mathrm{FCR}=1.2+\frac{-0.51}{1+{\left(\frac{\mathrm{WT}}{50}\right)}^3}$$

(1)

In which WT is fish weight in grams.

In the FW + DW system, 50% of the commercial feed is substituted by pelletized duckweed. Feed contents that were used in the model calculations are presented in

Table 3. Parameters of commercial fish feed and fresh duckweed.

Parameter	Unit	Commercial Feed Tilapia		Commercial Feed Barramundi		Fresh Duckweed
		Value	Source	Value	Source	Value	Source
Dry matter content	mass fraction	90%	[37]	92%	[38]	5%	[39]
Protein content	mass fraction	32%	[37]	42%	[38]	38%	[40]
N content	mass fraction	16% of PC 1	[41]	16% of PC	[42]	8.75%	[43]
P content	mass fraction	1.5%	[25]	1.1%	[38,42]	0.8%	[43]
COD content	g O2 g−1 DM	1.4	[42]	1.21	[38]	1.06	[44]

¹ Protein content.

.

Part of the feed is metabolized by the fish into biomass and another part is uneaten and ends up as solid sludge. Moreover, parts of N, P, and COD that are excreted by the fish can be in solid form as feces or insoluble form as urine. The parameter value of the fish metabolism follows from Dijkgraaf et al. (2019) [20] for both tilapia and barramundi. The fish metabolism values are crucial to calculate dilution and supplement requirements for HPS and methane production in the digester.

2.3. Plant Production System

Rice (Oryza sativa) and tomato (Solanum lycopersicum) is grown staggered in 1000 m² of the HPS facilities, respectively, with constant yield. Rice is grown in troughs with a water depth of 0.15 m and harvested every 120 days. Tomatoes are grown in 0.5 m deep troughs with a cycle length of 145 days, as described by Dijkgraaf et al. (2019) [20]. The water temperature is assumed to be constant. Duckweed is cultivated in a non-aerated pond because fast water current limits duckweed cultivation [45]. The DCS has a depth of 0.5 m to prevent problems during harvesting duckweed [46] and has a total area of 300 m².

In the model calculation, maximal yields of rice, tomato, and duckweed are presumed, because Indonesia’s climate conditions are optimal for their growth [47,48]. Climatological data of Jakarta (6°12′ S, 106°50′ E) were obtained from WorldWeatherOnline [49] and are presented in Appendix A,

Table A2. Indonesia weather data, compiled from weather data of ten years (2009–2018).

	Max Temperature [49]	Min Temperature [49]	RHmax [49]	RHmin [49]	RH [49]	Solar Radiation (Rs) [57]	ET0 *
	°C	°C	%	%	%	MJ/m2/day	mm/day
January	29	25	83.93	68.67	76.3	9.0	5.56
February	28.9	24.9	86.35	70.65	78.5	9.7	5.50
March	29.8	25	85.14	69.66	77.4	12.0	5.78
April	30.2	25.3	84.15	68.85	76.5	12.7	5.21
May	30.5	25.4	82.39	67.41	74.9	11.8	5.02
June	30	25.1	82.5	67.5	75	11.4	4.06
July	29.9	24.2	81.62	66.78	74.2	12.5	4.39
August	30.4	24.1	79.09	64.71	71.9	14.6	5.10
September	31	25	76.01	62.19	69.1	16.1	5.86
October	31.6	25.8	75.02	61.38	68.2	14.9	6.14
November	31.1	26.2	78.54	64.26	71.4	12.5	5.70
December	30.3	25.7	79.64	65.16	72.4	12.7	5.77

* ET₀ is reference evapotranspiration.

. In the system, rice yield is assumed to be 2.2 kg/m², which is around 4 times the yield of rice cultivated in the conventional way in Indonesia. Rice cultivated in HPS can even lead to yields that are five times higher than rice cultivated in an open field [50]. Duckweed density is maintained at levels higher than 0.6 kg wet biomass/m² to prevent algal blooms and lower than 1.2 kg wet biomass/m² to avoid growth inhibition [39]. Relative growth rate of duckweed is 0.3 day⁻¹ [43]. Tomato yields with EC between 0 to 2.2 dS/m are assumed to be constant at 16 kg/m² [20]. A yield reduction of 10% occurs for every 1 dS/m above the limit [51]. Except yield, parameters related to tomato growth follow from Dijkgraaf et al. (2019) [20].

Nutrient concentrations in the HPS nutrient solution are always maintained at optimal levels by dilution or adding supplementary nutrients to ensure maximal yield. Previous research suggests significant variation of the optimal nutrient concentrations for the three plants. Nutrient concentrations used in the model are presented in

Table 4. N and P concentration required in the nutrient solution for rice, tomato, and duckweed production.

Nutrients	Rice	Tomato	Duckweed
NO3-N	45–98 mg L−1	100–200 mg L−1	10–30 mg L−1
P	10–20 mg L−1	32–48 mg L−1	5–8 mg L−1

.

The evapotranspiration rate for HPS and DCS follows from Dijkgraaf et al. (2019) [20]. The crop coefficient (Kc) of rice changes over time and is based on the plant life stage, which is 0.6 in the initial, 1.2 in the middle, and 0.75 in the late stage [52]. However, for duckweed, only one value of Kc is used due to the short life cycle.

In the HPS, evaporated water is replenished by pumping water from the RAS or groundwater when the nutrient solution in RAS reaches its limits in either N or P. In the DCS, water is replenished by groundwater since lower nutrient concentrations are needed in the DCS than the HPS. However, if the volume of DCS is appropriately chosen, dilution in the DCS might be absent.

3. Model Description

The model is built from mass, water, and energy balances in the different system elements (Figure 1). In general, equations, assumptions, and order of calculation used by Dijkgraaf et al. (2019) [20] for the balances are also valid in FW − DW, FW tomato, and BW systems. However, some adjustments of their equations were applied (Supplementary Materials, Table S1). In the FW + DW case, the model needs to be extended. Mass balances of water, nutrients (N and P), and energy of FW + DW for the RAS, HPS, and DCS are shown in Appendix B, Equations (A1)–(A14).

Duckweed grown in the DCS are harvested frequently, dried, and pelletized. When duckweed biomass reaches its maximum density, 45% of the biomass is harvested to maintain the optimum growth potential. Equations for the overall mass balance and daily production of duckweed biomass can be found in Appendix B, Equations (A15)–(A16). The amount of pelletized duckweed available, and to be used as fish feed, is defined in Appendix B, Equation (A17).

All subsystems are assumed to be homogenous, making the state variables only a function of time. The resulting difference equations were solved in Microsoft Excel^TM using the forward Euler method, with a step size of one day.

3.1. System Performance Analysis

Simulated model outputs were defined and evaluated for 3000 days to compare the performances of the four systems. System performance indicators used are energy consumption, water use efficiency (WUE), nutrient use efficiency (NUE), breakeven time (BET), and net present value (NPV). To calculate the energy consumption, the amount of external energy bought from the national electricity company was used. Equations used to calculate WUE, NUE, and NPV are shown in

Table 5. Equations used for system performance and sensitivity analysis, including water use efficiency (WUE), nutrient use efficiency (NUE), net present value (NPV), and normalized sensitivity coefficient (S_yx).

	Equations	Descriptions
WUE	$\frac{\sum \mathrm{input}-\sum \mathrm{waste}}{\sum \mathrm{input}}$	∑input is total amount of groundwater; ∑waste is water discharged from HPS dilution, DCS dilution, and digester waste sludge.
NUE	$\frac{\sum \mathrm{input}-\sum \mathrm{waste}}{\sum \mathrm{input}}$	∑input is the total amount of nutrients used; ∑waste is total amount of nutrients disposed from HPS dilution, DCS dilution, and digester waste sludge.
NPV	${\displaystyle {\displaystyle \sum }_{\mathrm{k}=1}^{\mathrm{n}}}\frac{{\mathrm{R}}_{\mathrm{k}}}{{\left(1+\mathrm{i}\right)}^{\mathrm{k}}}$	${\mathrm{R}}_{\mathrm{k}}$(€) is net inflow minus outflow at time instant $\mathrm{k}=1,\dots , \mathrm{n}$ with $\mathrm{n}=8$; i is the rate of return (assumed to be 5%).
Syx	$\frac{\partial \mathrm{y}}{\partial \mathrm{x}} . \frac{\overline{\mathrm{x}}}{\overline{\mathrm{y}}}$	y and x are a specific model output and parameter, respectively; $\overline{\mathrm{x}}$ and $\overline{\mathrm{y}}$ are nominal value of the model output and parameter, respectively.

. NPV is calculated for eight years. The composition and price used for calculating the net cash flows are shown in Appendix A,

Table A3. Composition of revenues and expenses.

Revenues	Price (EUR)
Nile tilapia	1.5/kg
Rice	0.77/kg
Barramundi	4.8/kg
Tomato	0.44/kg
Expenses
Variable costs
Energy	0.1/kWh
Minimum wage	254.2/month
Land	2.67/m2/year
Feed	0.54/kg
N supplementation	0.58/kg
P supplementation	0.74/kg
Rice seed	1.92/kg
Tomato seed	0.89/g
Nile tilapia juvenile	0.05/ individual
Barramundi juvenile	0.19/individual
Investment costs
RAS 1	2/kg fish
HPS 2	8.49/m2
DCS 3	1.6/m2
ANA	600/unit

¹ The investment cost of the RAS is derived from Timmons and Ebeling (2010) [36]. The value used in this model is only 75% of the total equipment and facilities costs because it was assumed that the technology used in IndonesiAP is lower than in the mentioned study. Furthermore, the study of Timmons and Ebeling included heating facilities, which are not necessary for Indonesia. ² The investment cost of the HPS is gathered from Bengkel Mimpi which gave information of the investment cost for both aquaculture and rice hydroponic facilities. In the current study, it is assumed that total investment cost required for the HPS is 40% of the total investment cost used in Bengkel Mimpi. ³ The cost of investment for DCS is from Skillicorn et al. (1993) [58]. Since the value in the report is for year 1993, future value of the cost for the year 2019 was calculated assuming a rate of 5%.

. NPV is calculated per year, thus index k in Table 4 runs from 1 to 8. In addition, two types of costs, i.e., variable and investment cost, were considered. The investment cost is spent only in the first year.

3.2. Sensitivity Analysis of System Design Parameters

System design optimization aimed to maximize the NUE and WUE of the system under the constraint that the system is economically feasible. The optimization was done by manually varying the value of the most sensitive parameters in the model. The most sensitive parameters were determined by a sensitivity analysis of the design parameters, including protein and P content of the commercial feed, fish density, RAS volume, HPS area, and DCS area, duckweed harvesting strategy related to biomass density in the DCS, percentage of duckweed in fish feed, land price, labor cost, selling price of products, and investment cost. These parameters were increased and decreased by 50% from the initial value (later called nominal value). A normalized (non-dimensional) sensitivity coefficient (S_y) was used for a direct comparison (Table 5).

4. Results

The model calculations show that an individual barramundi fish requires more feed and gains more biomass (around 300 g higher) than Nile tilapia when harvested after 200 days of cultivation (Figure 2). Moreover, a different pattern in the feed of tilapia and barramundi is observed in Figure 2. A change in the feed requirement of tilapia occurs from around day 110 to 150 but does not occur in feed of barramundi.

The FW + DW system aimed to substitute 50% of commercial feed with pelletized duckweed produced in a 300 m² DCS. However, a 300 m² DCS is not enough to cover all duckweed required in the RAS, as shown in Figure 3. Duckweed can substitute 50% of the daily feed requirement until approximately day 160. After that day, duckweed pellet production can only cover the feed requirement for around ten days every 50 days. A DCS of 1350 m² is required to cover 50% of tilapia total feed (Appendix C, Figure A1). Another function of DCS is to capture excess nutrients from HPS. The model calculations found that a 300 m² DCS is enough to capture all excess N from the HPS and sufficient to support duckweed growth without additional fertilizer from around day 180 onwards. Additionally, the DCS can increase the NUE of P, NUE of N, and WUE by around 8%, 2%, and 5%, respectively (Appendix C, Figure A2, Figure A3 and Figure A4). However, such size is not enough to capture all P excess from the HPS (Figure 4). Therefore, dilution is still needed in the DCS due to high concentration of P (Figure 4).

N and P concentrations in the FW + DW, FW − DW, and BW system have a pattern of 365 days (Figure 4 and Figure 5), which follows the evaporation trend in HPS (Appendix C, Figure A5). Figure 4 and Figure 5 also show that N concentration in RAS is always higher than P. Compared to the FW system, N and P concentration in the RAS unit of the BW system is higher. Further, N and P concentration in the RAS also keeps decreasing for some days after harvesting.

Notice from Figure 4 and Figure 5 that in the HPS of the FW systems, additional N and P are supplied until approximately day 150 and 120, respectively. After these days, nutrients supplied from the wastewater of the ADU and RAS water is enough to support plant growth without additional supplementation. Without supplements added to the system, N and P concentration in HPS increases until around day 200. At that time, P concentration achieves its threshold for potential rice growth. Consequently, dilution is needed, hence lowering both N and P concentrations in the HPS.

Compared to the FW system which coupled with rice, the BW system which coupled with tomato shows a substantially improvement of NUE and WUE as it reaches more than 95% in the BW system (Appendix C, Figure A2, Figure A3 and Figure A4). Nutrient supplementation in the BW system only occurs in the first 150 days. Dilution of the BW system is not needed. For the FW system, a 1000 m² area of HPS growing rice is inadequate to get high use efficiencies of water and nutrients, and especially P. The P concentration in HPS always reaches the maximum threshold at approximately day 200 onwards (Figure 4, left panel), indicating the need for dilution.

Evaluation of NPV (Figure 6) indicates that building an IndonesiAP FW − DW system in Jakarta is not economically feasible. DCS is expected to improve the profitability of the FW system by reducing feed cost. Indeed, on the one hand, duckweed pellet substitution can lower the feed cost by around EUR 2000 for a system running for 3000 days (Appendix A,

Table A4. Total revenue and expenses of both FW − DW and FW + DW in the IndonesiAP system after 3000 days of operation. Brackets indicates minus values.

	FW – DW12345678(EUR)		FW + DW(EUR)	(FW − DW) − (FW + DW)12345678(EUR)
Revenue
Nile tilapia	42,251		42,251	0
Rice	43,951		43,951	0
Total revenue	86,202		86,202	0
Variable expenses
Tilapia seed	307		307	0
Rice seed	51		51	0
Feed	18,242		16,139	2102
Supplementation	16		18	(3)
N supplementation	12		14	(1)
P supplementation	3		5	(1)
Energy	20,481		20,509	(28)
Labour	11,757		15,117	(3359)
Land	23,020		29,597	(6577)
Total variable expenses	73,874		81,738	(7864)
Fixed expenses
RAS	18,000		18,000	0
HPS	8492		8492	0
ANA	600		600	0
DCS	0		480	(480)
Total fixed expenses	27,572		27,092	(480)
Total expenses	100,966		109,310	(8344)

). On the other hand, this design increases the expenses by about EUR 10,000 (Appendix A, Table A4) to cover additional energy, land, labor, and duckweed pond installation. In summary, the FW − DW exceeds the FW + DW system in terms of economy, although neither system is feasible, based on the nominal values used in the model. Notice from Figure 6 that the BW system improves both NPV and BET of the system significantly compared to the FW system. However, if rice in the HPS of the FW system is substituted by tomato (FW tomato), the FW system becomes more interesting than the BW system in terms of profitability.

The performance of the FW + DW and FW − DW system is suboptimal. These systems can be improved by varying the value of some of the most sensitive design parameters, namely fish density, feed P content, rice area, and duckweed area (Appendix C, Figure A9). A best-case scenario is achieved by lowering feed P content to 0.6% and doubling the fish density (

Table 6. Nominal and best case of FW + DW (columns 3–4) and FW − DW system (columns 5–6).

System Parameters	Unit	Nominal FW − DW	Best Case FW − DW	Nominal FW + DW	Best Case FW + DW
HPS area	m2	1000	1300	1000	1000
Feed P content	%	1.5	0.6	1.5	0.6
Fish density	kg/m3	40	80	40	80
DCS area	m2			300	300
NUE N		85%	94%	89%	91%
NUE P		78%	94%	88%	91%
WUE		68%	96%	80%	81%
Energy consumption	kWh	2.05 × 10−5	1.97 × 105	2.05 × 10−5	1.97 × 105
NPV	€	(18,544)	905	(25,224)	(5211)

). Moreover, an expansion of 300 m² HPS is required in the FW − DW system. Although the nutrient and water efficiency can be increased, the NPV is still negative in FW + DW and low in FW − DW after 3000 days of production.

5. Discussion

This study explored several virtual aquaponics systems in Indonesia and evaluated NUE, WUE, and economic analysis of the proposed system. Different patterns of the daily feed requirement of tilapia and barramundi were found (Figure 2). This difference is a result of constant FCR in Nile tilapia before day 110 and after 150. Between the mentioned days, the FCR increases from around 0.7 to 1.3 (Appendix C, Figure A6). Consequently, the increase in feed requirement also rises. In contrast, the FCR of barramundi increases during the first 50 days and then remains constant at 1.2 (Appendix C, Figure A6). Hence, the feed requirement observed in Figure 2 for barramundi is almost linear.

N and P concentrations in the RAS are determined by the composition of the fish feed. N is always higher than P because the feed added to the RAS has more N than P (9–16% N and 0.8–1.5% P, Table 3), and fish soluble excretion of TAN is higher than P [53]. N and P concentration in the RAS unit of the BW system is higher than in the FW systems, because the barramundi feed contains more N (Table 3) and the fish excretes more soluble N and P than tilapia [38,53]. A higher N excreted by barramundi indicates that tilapia uses N more efficiently than barramundi. Therefore, a higher biomass gain is expected in tilapia, if the feed protein concentration for both fish types is equal. However, the feed protein content in the BW system is 50%, while in the FW system it is only 32%. Hence, it is still possible for barramundi to gain more weight than tilapia (Figure 2).

Furthermore, N and P concentration in the RAS keeps decreasing for some days after harvesting, when a new batch of fingerling batch is added to the tank. Hence, a lower feed rate is required after the harvesting day. Consequently, nutrient input through the feed at that time has a lower concentration than the discharged nutrients through water pumped from the RAS to the HPS to cover evapotranspiration. Around 20 days after harvesting, the feed consumption rate of the fish increases due to the growth of the fish. As a result, nutrients concentration in the RAS increases again until the next harvesting day.

The pattern of N and P in the RAS are in line with the evapotranspiration in the HPS, because RAS water is pumped to the HPS to cover the evapotranspiration of the plants. Water pumped from the RAS to the HPS is replenished by groundwater, which has lower nutrient concentrations than RAS water. Consequently, nutrient concentrations in the RAS drop at high evapotranspiration rates.

Dilution of the HPS is needed due to the excess of P, not N, since N concentration in the HPS never hit its thresholds, while P is at its upper limit for most of the time (Figure 4 and Figure 5). Although N concentration in the RAS is higher than P, the mobilization rate of P in the ADU is almost four times higher than the mobilization rate of N. This explains why P is more dominant than N in the nutrient’s solution of the HPS. To minimize dilution of HPS, the rice area should be increased by at least 300 m², which subsequently will increase NUE and WUE to around 90%. A higher NUE and WUE observed in the BW system can be explained by the fact that the N and P levels and ranges for tomato are larger than for rice (Table 4). For instance, rice requires a P concentration between 10–19 mg/L, while tomato can tolerate a P concentration of 32–48 mg/L. Since P is the leading cause of dilution of the system, a higher upper limit of P will sharply reduce the need for dilution.

A DCS can increase NUE of the FW + DW system, in which the effect on NUE P is more substantial than NUE N. Installing DCS means adding a pool to capture the excess P from HPS to produce duckweed biomass, which is subsequently used as fish feed in the RAS. By doing so, the dilution requirements of HPS and feed costs are reduced. In addition, WUE also depends on the size of the DCS. A DCS area of around 175 m² is already enough to make FW + DW exceed the FW − DW system in terms of WUE. On the other hand, feed cost reduction is not enough to cover the extra costs spent on building and operating the DCS installation, such as land, labor, and investment costs. In this study, installing an FW + DW system in Jakarta is not recommended due to high land prices. The total amount spent on land and labor is around five times higher than the reduction of feed costs.

The choice of the size of DCS depends on the system’s objective, whether to fulfil feed requirements or to make optimal use of water and nutrients. Based on our model calculation and if the goal is to maximize NUE and WUE, doubling the DCS area is already enough to make the efficiencies higher than 90%. If the goal is to fulfil 50% of the total feed requirement, the area must be at least 4.5 times higher than the nominal value of 300 m². If the aim is to maximize NUE and WUE by eliminating the dilution of the system, an area of around 1000 m² is enough. Increasing the DCS means more duckweed is available to take up the nutrients discharged from the AP; hence, at a certain point dilution is not necessary anymore.

In the study of the FW + DW and FW − DW systems with respect to maximizing NUE and WUE, four plausible strategies were suggested, namely increasing rice and duckweed area, and lowering fish density and feed P content. From these options, reducing fish density is the least preferred strategy, because it can significantly reduce the NPV while insignificantly improve NUE and WUE. The most favorable strategy is to supply fish with a lower feed P concentration, since it dominates NUE and WUE without affecting NPV. P is also an exhaustible resource and the leading cause of nutrients dilution in the IndonesiAP system. However, it should be noted that the feed P content must still be in the acceptable range to grow optimally. Based on the model calculations, feed P as low as 1% is still useful to improve NUE and WUE of the system (Appendix C, Figure A7). This value is also still acceptable to Nile tilapia because the nutrient composition of typical tilapia feeds has a minimum available phosphorus content of 0.6% [35]. Furthermore, upscaling the HPS area gives better results on the NUE and WUE than upscaling the DCS area, since rice requires higher N and P levels than duckweed. Although both strategies negatively affect the system’s profitability, the rice plantation area has a less negative effect. The sensitivity coefficient of NUE and WUE with respect to these parameters might turn into positive values if the land price is lowered. Hence, at a lower land price, enlarging the HPS area can be chosen to increase NUE and WUE of the system in addition to lowering feed P content.

NPV becomes the main concern in the FW + DW system because installing DCS under the given nominal values is discouraged due to the high expenses. Based on the sensitivity analysis (Appendix C, Figure A2), NPV is, after fish density, most sensitive to the land price. However, in terms of NPV, even if the land price is only 10% of the nominal value, installing DCS to reduce 50% of feed required will never compete with the standard aquaponics system, in which all feed is bought from suppliers (Appendix A,

Table A5. NPV of optimized scenario of FW − DW and FW + DW systems (rice area of 1300 m², feed P content of 0.6%, and fish density of 80 kg/m³) if land price, salary cost, feed price, and investment cost of the scenarios is reduced.

	FW − DW	FW + DW
	NPV (EUR)	NPV (EUR)
Nominal Value	905	(5804)
Land price 50%	12,569	8455
Land price 10%	21,900	19,860
Salary cost 50%	6862	1481
Feed price 50%	15,129	7595
HPS investment 50%	6425	(281)
RAS investment 50%	9905	3199
DCS investment 50%		(5561)
Land price 10% and salary cost 50%	2419	1945
Land price 10%, salary cost 50%, and feed price 150%	13,634	13,745

). Further, decreasing land prices becomes more logical if the minimum wage also decreases. The minimum wage in some provinces in Indonesia can reach as low as 50% of Jakarta’s minimum salary [54]. Reducing both land price and labor cost to only 10% and 50% of the nominal value, respectively, still cannot exceed the FW − DW system in terms of NPV and BET. The FW + DW scenario starts to be more favored than FW − DW when the feed price increases by about 50% from the nominal value. However, if the duckweed maximum crop density can be increased up to the number used by Frederic et al. (2006) [55], reducing land prices to only 10% of the nominal value is enough to make a DCS extension profitable.

The economic feasibility of the aquaponics system is considerably improved when the BW system is chosen. However, substituting rice with tomato in the FW system (FW tomato) results in higher NPV than the NPV of the BW system. Thus, based on the results from the sensitivity analysis, increasing tomato production seems to be a promising solution to increase the economic feasibility of both FW and BW tomato systems without harassing the environment. Expanding the HPS area of tomato by 35% can make the NPV of the BW system surpass the nominal FW tomato system (Appendix C, Figure A8). In addition to increasing the HPS area, increasing fish density is also a possible approach to improving the BW system’s profitability without significantly decreasing NUE and WUE of the system. However, increasing fish density in the BW system will never surpass the NPV of the FW tomato system because the sensitivity coefficient of fish density in FW is higher than in the BW system (Appendix C, Figure A9and Figure A11).

A possible way to make the BW system economically more feasible than the FW tomato system is by increasing the price of tomato grown in brackish water. Tomatoes cultivated in water with higher salinity contain higher dry matter content, resulting in a better taste than tomatoes grown in fresh water [50]. Hence, it seems to be realistic to evaluate a higher selling price of tomato produced in the BW system. An increase of around 20% of the selling price already makes the BW system an economically more feasible system than the FW tomato system (Appendix C, Figure A8). If there is also a chance to raise the barramundi selling price, a 20% increase makes the NPV of the BW system exceed the NPV of the FW tomato system (Appendix C, Figure A8).

This study only covers DAPS, in which water is only transported from the RAS to HPS and the wastewater from HPS or DCS is discharged from the system. In addition to DAPS, an on-demand coupled aquaponics system configuration with a second connection, i.e., with a regained water flow from HPS to RAS [18], is worth evaluating in the future, especially in countries where water is scarce. Hence, water requirements can be minimized, too. Finally, although the results indicate a potential of introducing aquaponics systems in Indonesia, the model needs to be calibrated and validated, given data from a real plant, to ensure up-to-date model behavior.

6. Conclusions

This study examined the performance of fresh and brackish water aquaponics systems using a dynamic mathematical model. Our findings discourage the implementation of freshwater aquaponics systems with tilapia and rice in Jakarta, with or without a duckweed cultivation system, due to the high land prices. Substitution of rice by tomato improved the net present value of the systems. In addition to this, tomato has a wider range for the nutrients N and P than rice, which reduces the need for dilution. Minimizing the dilution of the hydroponic system results in higher nutrient and water use efficiencies. Adding duckweed in a freshwater aquaponics system with tomato is a potential solution to satisfy feed necessity in a system where protein source is scarce or expensive, and land and labor costs are low. However, it increases the complexity of such system. The performance of the brackish water system is comparable to the freshwater system that grows tomato in its hydroponic system.