Produce via Georgia, Upper Lars up 13% and flow meters: what reaches a greenhouse or winery

In one week, two groups of news of very different scale came out. The first is about freight: Russian Railways is looking into a route for vegetables and fruit from Turkey and Azerbaijan to Russia through Georgia, and in September the Upper Lars crossing handled 13% more cargo than a year earlier. The second is about hardware: a clamp-on ultrasonic flow meter, fail-safe valve actuators and a holding company with a single digital ecosystem. To a greenhouse or winery owner in Georgia, the first group says that Turkish produce will travel past them in volume, and the second says that water, heat and equipment condition can be measured without stopping production. Of all this, only what lands on the server reaches the farm.

Freight through Georgia

On October 23, Alta-Soft quoted Vadim Mikhailov, first deputy CEO of Russian Railways: a land route is being worked out for Russian retailers to bring fruit and vegetables from Turkey and Azerbaijan in transit through Georgia.

The same day, Alta-Soft passed on figures from the North Ossetian customs office: 503 thousand tons went through Upper Lars in September, 13% more than in September 2025, and 3.4 million tons over nine months. Turkey leads by cargo: fruit and vegetables, consumer goods, equipment, plastics and semi-finished products for food production.

I'll put this carefully: these are transit numbers, and they don't describe the local market. But the flow is growing, and a Georgian greenhouse will have a hard time competing with Turkish tomatoes on price. What's left is competing on precision: knowing how much water and how many kilowatts go into a kilogram of harvest, and where a batch gets lost.

A flow meter, an actuator and an ecosystem

On October 23, Engineer Live covered the PCE-TDS 200 from PCE Instruments, a portable clamp-on ultrasonic flow meter. Two sensors are fixed to the outside of the pipe with clamps and gel; the device measures the difference in signal travel time with the flow and against it, and calculates velocity, flow rate and total volume. It covers pipes from DN 15 to DN 6000 and media from minus 30 to plus 160 °C. There's a built-in data logger with USB export, and with temperature sensors added the device calculates the thermal power of the loop. Nobody cuts the pipe, and production doesn't stop.

The same outlet on October 19 ran a Rotork piece about Skilmatic SI electro-hydraulic actuators for tank farms and terminals. If the signal or power is lost, the actuator moves the valve to a safe position on its own: open, closed or as it was. The SI3 and SI4 models are suitable for SIL 2 and SIL 3 systems, work from minus 50 to plus 70 °C, and a built-in logger stores up to 3,000 events.

On October 13, CNews reported that the Agrosila holding is developing a single digital ecosystem that ties together its production, technology and corporate systems, and is rolling out AI tools. I haven't seen anything beyond the announcement, so I can't say what exactly they implemented or with what result.

Example: a winery in October

Take a winery where the cooling runs at its limit in October. Once a day the winemaker writes tank temperatures down in a notebook. The cooling pump on one tank stopped during the night, and nobody knew until morning. The batch is fine, but its history now rests on what people remember. A flow meter on the cooling pipe and a logger in the pump actuator would both have recorded these events. Only the records would sit inside the devices, and the winemaker would see them when he came by with a laptop and a USB cable.

So a monthly USB export gives you an archive, and that's all. Monitoring starts when data reaches the server on its own, on a schedule.

How it works at the farm I work with

Since 2023 I have been responsible for the server, network and database of a hydroponic farm in the US. The equipment is run by Raspberry Pi based controllers: irrigation, pumps and other equipment all work through them. There are now 2 sites and 25 controllers on one network. The site networks sit on private IP addresses with no public address, so I set up reverse proxies, NAT, a firewall and a VPN, and the server got access to every controller. The server is a Debian virtual machine in Google Cloud, with MariaDB on the same machine. Data comes from the controllers into the database, and commands go back to the equipment.

The hardest part was the controllers' closed system: there was no REST API, and the only option offered was local sockets. I figured out how that interface works and built the integration on top of it. Another team does the automation logic and the interface; I handle what sits underneath: the network, the server, the database, updates and support.

For greenhouses and wineries in Georgia I offer the same approach: start with one site and one working task. For example, a temperature history for each tank and an alert when the cooling stops. I don't have savings figures for a winery, and I won't promise any before I've looked at the site.

What I would do

Start with a list of equipment: what can already send data and what stays silent. For the PCE-TDS 200 the article mentions only USB, so any outside interface has to be checked in the documentation.

Then one server and one database where data arrives on a schedule. At the farm that's Debian and MariaDB, and for a winery I'd start with the same.

If the sites are on private IP addresses, set up a VPN and a reverse proxy before the first chart appears.

Keep the first screen simple: temperature by tank, water flow by line, pump status and one Telegram alert when something goes out of range.

I wrote about why a system needs a shared center in the article The server: data, access and process execution. Earlier I looked at keeping analytics next to the working archive of industrial data. On the same topic: models on devices across a fleet of sites and rejected loads and port bookings. Services: greenhouse automation, irrigation and pump control, systems for wineries and equipment monitoring. Project: a hydroponic farm in the US. If you have equipment that writes data only to itself, describe what already works and what you can't see. Discuss your task.