Modular BSF Facility Design: From One Tonne a Day to a Thousand

What This Page Answers

Anyone evaluating a BSFL operation faces two facility-design questions that determine most of the capital and operating cost. First: what are the five functional zones every BSF facility requires regardless of scale, and what are the non-negotiable specifications for each? Second: how does the design change across throughput tiers, and where are the automation thresholds that make each tier financially distinct from the one below it?

This page answers both questions with the level of specificity needed to scope a facility, commission an engineering study, and avoid the capital sequencing errors that have defined the failure history of the European BSFL sector. For background on why the underlying biology makes this species commercially viable at all, and for the feedstock sourcing strategies that make any facility viable before the building is designed, see the sibling pages in this cluster.

The failure autopsy on AgriProtein and several EU pilot facilities from 2018-2021 consistently identifies the same sequence: facility sized for future scale, feedstock supply at 10-20% of nameplate capacity at launch, opex bleeding cash for 18-36 months before revenue catches throughput, equity exhausted before the operating curve inflects. Modular design with a committed feedstock-first approach is the countermeasure. See food waste feedstock sourcing for how to secure that feedstock before building.

The Mechanism: Five Facility Zones and What Each Requires

Every BSF facility, whether a 1 TPD farm shed operation or a 250 TPD industrial plant, passes material through five functional zones in sequence. The zones are not negotiable: skip one and either biology fails or product quality drops below the specification that buyers require.

Zone 2 (rearing) is the rate-limiting zone. Stocking density, ventilation, and temperature control determine yield per square metre of rearing floor. Over-stocking generates larval heat and ammonia that reduce growth rate by 15-30% and increase mortality. Under-stocking wastes capital and floor space. The operating target is 5-10 kg of wet larvae per square metre of rearing tray surface at harvest, depending on feedstock composition. Protein-rich feedstock (bakery waste, brewery spent grain) supports the higher end; lower-energy feedstock (vegetable washing water) supports the lower end.

Zone 4 (processing) is the most capital-intensive zone per unit of output because it requires heat, pressing, and drying equipment that scales discontinuously. A 1 TPD facility can use a second-hand olive press and a batch dryer; a 100 TPD facility needs a continuous screw press line and a drum dryer with heat recovery. The jump from batch to continuous processing happens somewhere between 5 and 15 TPD depending on labour cost in the operating country. In Germany or the Netherlands, labour cost justifies continuous processing at 8-10 TPD. In Southeast Asia, manual batch processing can be economical at 20 TPD.

Zone 3 (harvest and separation) has an elegant engineering shortcut at scale: prepupae self-harvest. As larvae approach the prepupal stage at day 12-14, they develop a wandering behaviour and attempt to migrate away from the feeding substrate to find a dry pupation site. A facility that channels this migration through a ramp into a collection bin harvests 60-75% of larvae without any mechanical sieving, reducing damage and labour cost. This self-harvest effect is well-documented (Tomberlin et al. 2002) and is the design principle behind continuous-flow rearing systems at 50 TPD and above. See also the conversion math for how harvest efficiency affects revenue per tonne of input.

The Numbers: Capital Cost, Footprint, and Automation Thresholds

Capital cost per tonne of daily feedstock capacity falls as throughput increases, but not linearly. The drop is steep from 1 TPD to 10 TPD as batch equipment is replaced by continuous equipment. From 10 to 100 TPD the drop continues but flattens. Above 100 TPD, additional scale delivers smaller cost improvements per tonne and increasing integration complexity risk.

The automation threshold is the key design decision. At 1-5 TPD, manual tray handling by two to three workers covers the rearing zone at acceptable labour cost. At 10-15 TPD, manual tray handling requires 8-12 workers per shift for rearing alone, and labour cost begins to exceed 40% of opex in Western European markets. Semi-automated conveyor-fed rearing trays and automated sieving at harvest reduce labour need to 3-5 workers per shift at 10-15 TPD. The capital cost of semi-automation at this scale is 300,000-600,000 EUR for the rearing and harvest zones, paid back in labour savings within 18-36 months at European wage rates.

The footprint calculation is straightforward once throughput is fixed. Rearing floor area in square metres equals daily dry matter input in tonnes divided by 0.05 (the maximum stocking density per square metre of tray surface). For 10 TPD wet feedstock at 75% moisture, dry matter input is 2.5 tonnes per day, requiring 50 square metres of active tray surface. With a 14-day batch cycle and 2-3 metre stacking height, total rearing room footprint including aisle space is approximately 200-400 square metres. Add processing, storage, and ancillary space and the total facility footprint for 10 TPD is 800-1,500 square metres, which fits a standard industrial unit in most European business parks.

The Practitioner View: Protix and the Modular Scale Path

Protix in Bergen op Zoom, Netherlands, is the closest thing the European BSFL sector has to a validated industrial template. Founded in 2009, Protix spent the first six years at research and pilot scale under 500 square metres of rearing space, validating both feedstock sourcing (food industry side streams from the Dutch food cluster, one of the densest in Europe) and offtake (aquaculture feed manufacturers including Skretting). The industrial facility at 14,000 square metres was not designed until feedstock and offtake contracts were in place. This sequencing is why Protix survived and many contemporaries did not (vault_atom_TBD: Protix disclosures 2019-2023).

The Buhler partnership is the automation reference point. Buhler, the Swiss food and feed processing equipment manufacturer, co-developed the continuous rearing and processing line at Bergen op Zoom, establishing what is now the de facto specification for industrial BSF automation. The system uses sensor-monitored rearing trays, automated feeding conveyors, and a continuous harvest line that captures the self-harvesting prepupal migration with 70-80% efficiency before mechanical sieving cleans the remainder. At 250 TPD, the rearing zone requires approximately 50-80 workers per day across three shifts for monitoring, tray management, and quality control, against 120-150 workers that a manual equivalent would require. Labour cost per tonne of output is 40-50% lower than manual at industrial scale.

The contrast with AgriProtein is instructive. AgriProtein built the world's largest fly facility at the time in South Africa and the UK, sized for 200-500 TPD, before securing feedstock at even 20% of nameplate capacity. Operating at 10-15% capacity utilisation for 18-30 months before capital exhaustion, the unit economics never inflected. The technology worked. The business sequencing did not. The lesson for any new operator: size the first module to the feedstock you have under signed contract today, not the feedstock you expect to have in three years. Build module two when you need it.

For operations in the agricultural automation track, BSF facility automation is one of the most clearly defined ROI cases in agri-food: the biology is deterministic, the production cycle is 14 days, and sensor data (temperature, humidity, CO2, larval mass) maps directly to controllable actuators (ventilation, feeding rate, water spray). This makes BSF facilities well-suited to process automation investment, where payback periods of 2-4 years are common at mid-scale.

Where Facility Design Fits in the BSF Investment Decision

Facility design is step three in the investment decision, not step one. Step one is feedstock security (signed contracts for a defined volume, composition, and delivery schedule). Step two is offtake security (a buyer for protein meal and frass, at a price that covers operating cost). Only after both are in place does facility design earn the engineering investment. The mistake of the 2016-2021 European cohort was treating facility design as a fundraising artefact: build an impressive factory, then find the feedstock and buyers. The facilities that have survived reversed this order.

This page connects to two other spokes that define the investment context. Food waste feedstock sourcing covers how to structure the contracts and due diligence that make feedstock security real rather than assumed. The BSFL vs soy economics page covers the offtake price dynamics: specifically, when does the BSFL protein price justify building against soy as the incumbent, and which buyer segments are decoupled from the soy price correlation that affects commodity aquaculture and poultry feed.

The modular design principle also connects to how a BSFL operation integrates with the poultry feed and fish feed supply chains. A 10 TPD facility produces roughly 700-1,000 kg of dried protein meal per day. At 2,100 EUR per tonne, that is 1,470-2,100 EUR per day in protein meal revenue before frass. To put this in context: a 10,000-bird laying hen operation consuming 100 g of feed per bird per day at 15% BSFL inclusion needs 150 kg of BSFL meal per day. A single 10 TPD facility can supply 5-7 such poultry operations simultaneously, creating a local circular loop that does not require freight logistics across multiple regions. This is the scale where the BSF-poultry-frass-soil cycle is practically achievable without industrial infrastructure.

The forward path for facility design is vertical integration and heat recovery. Facilities that co-locate with a food processing plant (as InnovaFeed does with Tereos in Nesle) receive feedstock at minimal transport cost, use waste heat from the host plant for rearing climate control and product drying, and deliver frass to nearby agricultural land. This configuration reduces energy opex by 20-35% against a standalone facility and eliminates feedstock transport entirely. It is the design direction that breaks the remaining economic barrier to viable operations at 5-20 TPD in European markets.