Cold Chain IoT in Agriculture: Cut Post-Harvest Loss
Tomatoes picked in Kaduna at dawn can be soft and unsellable by the time they cross Lagos traffic in an open truck. Nobody sees the moment the produce spoiled, so nobody can fix it. Cold chain IoT changes that: small sensors record temperature and humidity along the whole journey, and alerts reach a human while there is still time to act.
This article is a practical guide for farm owners, aggregators, processors and logistics managers who want to cut post-harvest loss without building a huge refrigerated network first.
The problem: losses nobody measures
Post-harvest loss is one of the biggest silent costs in African food systems. A recent Frontiers in Horticulture review discusses FAO-based estimates of large fruit and vegetable losses along the supply chain, and notes that figures vary and that interventions are often poorly targeted because reliable data is lacking. That last point is the heart of the matter: you cannot reduce what you cannot see.
In practice, losses come from a handful of causes:
- Heat exposure while waiting at farm gates, markets and loading bays.
- Unreliable power that turns cold rooms into warm rooms overnight.
- Slow handoffs between farm, aggregator, truck and retailer, with no record of who held the produce when.
- No proof for disputes, so buyers reject loads and suppliers cannot show the cause.
The solution: sensors, connectivity and alerts
A cold chain IoT system watches three places: the storage, the vehicle, and the handoff.
What to measure
| Asset | Sensors | Why it matters |
|---|---|---|
| Cold room or store | Temperature, humidity, door open, mains power status | Catches compressor failure and power cuts |
| Reefer or insulated truck | Temperature, GPS, door open | Proves conditions in transit |
| Crates and pallets | Low-cost data loggers or BLE tags | Tracks batches across handoffs |
| Ripening or drying room | Temperature, humidity, ethylene where relevant | Controls quality, not just cooling |
Connectivity that fits African sites
Not every farm has reliable cellular coverage. Use the right link for each site:
- GSM or LTE-M for trucks and sites with signal.
- LoRaWAN for sprawling farms and packhouses, with a local gateway on solar power. We cover this in our guide to IoT for Nigerian agriculture and climate monitoring.
- Store-and-forward loggers that record offline and upload when the truck reaches coverage.
A simple alert rule
The core logic is not complicated. A rule engine such as Node-RED, or a few lines in your backend, can drive it:
rule: cold_room_warning
when:
sensor: cold_room_temp
above: 8.0 # degrees C, set per crop
for: 15 minutes
then:
- whatsapp: store_manager
- sms: operations_lead
- escalate_after: 30 minutes
Thresholds must be set per product. Leafy greens, tomatoes, yams and fish all tolerate very different conditions, so involve your quality team.
Where AI adds value
Rules catch obvious breaches. Machine learning adds three things on top. Predictive analytics can flag a cold room that is drifting before it fails, based on compressor cycle patterns. Anomaly detection can spot a sensor that has stopped reporting or a door left open at night. And a shelf-life estimate can tell a buyer which batch to sell first. Start with rules, and add models once you have a few months of clean data.
Benefits
- Fewer spoiled loads because problems are caught in minutes, not at the destination.
- First-expired-first-out selling, driven by real conditions rather than guesses.
- Evidence for claims, with a time-stamped temperature record for each batch.
- Better access to premium buyers, since supermarkets and exporters increasingly ask for temperature records.
- Lower energy waste, because you can see which rooms run too cold or too often.
Not sure which part of your chain to sensor first? Talk to an expert and we will help you map the weak points.
Proof and examples
We want to be careful about claims. Temperature and humidity monitoring is mature, widely used technology, and the main risk in Africa is not the sensors but the power, connectivity and maintenance around them. A system that is not powered through outages, or that nobody responds to, achieves nothing.
For a feel of how we approach field-grade monitoring, see our StrikeClimate project, which shows our work on climate and environmental technology. The design lessons, solar-powered nodes, offline buffering and clear dashboards, transfer directly to a packhouse or truck fleet. If your produce moves by road, our article on AI route optimization for logistics in Lagos pairs well with this one, and so does AI crop disease detection at the farm end of the chain.
Pitfalls to plan for
Most failed pilots share the same causes. Sensors are installed but nobody owns the alerts. Batteries and gateways are not protected from outages. Thresholds are copied from a generic template rather than set per crop. And data sits in a dashboard that managers never open. Assign a named owner for each alert, test the escalation path monthly, and review the data in your regular operations meeting so that it changes decisions, not only reports.
Implementation roadmap
- Map the chain (week 1). Walk the route from farm to buyer and list where produce waits.
- Measure losses by stage (weeks 1 to 4). Even a manual count of rejected crates by stage gives a baseline.
- Pilot two or three sites (weeks 3 to 8). One cold room, one truck and one aggregation point is enough to learn.
- Add power backup. Solar plus battery for gateways and critical sensors, so data survives outages.
- Connect alerts to people. Use WhatsApp Business API messages with an escalation path. An alert nobody reads is wasted.
- Build the dashboard and batch reports (months 2 to 3). Show live status, history and exportable batch records.
- Scale and tune (months 4 to 6). Adjust thresholds, add sites and introduce predictive models.
Treat supplier and driver data with care. GPS traces of identifiable drivers are personal data under the NDPA 2023, so be clear with staff about what you collect and why.
FAQ
Do I need expensive refrigerated trucks?
No. Sensors help even with insulated trucks, ice or shaded loading, because they show where heat damage happens. That evidence tells you where cooling investment will pay back.
What happens when the power or network fails?
Good loggers store readings locally and upload later. Gateways should have battery or solar backup. Design for outages, because they will happen.
How much does it cost?
It depends on the number of sites, sensor type and connectivity. A pilot with a few sites is modest, and the larger cost is usually installation and process change. We can scope it with you.
Get started
If you aggregate, store, process or transport fresh produce, a small monitoring pilot can show you where your losses really happen. Book a free consultation or chat with us on WhatsApp and we will help you plan a pilot.