Features
Fine-tuning greenhouse energy controls

There’s a teaser lurking below the surface in greenhouse engineer Quade Digweed. For three years, he’s been teasing apart measurable components in light, heat, humidity and ventilation for greenhouse vegetable production. 

Now, his projects are coming together with insight into simultaneously saving money and fine-tuning climate control.  

Graduating in 2017 from the Lakehead University mechanical engineering program, Digweed was brought on staff by Agriculture and Agri-Food Canada as a junior greenhouse engineer-in-training at the Harrow Research and Development Centre in Kingsville, Ont.  

His mandate is to support controlled environment agriculture research and development. The focus is on understanding greenhouse microclimate, cover materials, and novel sensor applications.

Sites for the research are two new research greenhouses at Harrow and a cooperative partnership with Allegro Acres Inc., a nearby commercial greenhouse in Kingsville. Allegro Acres recently expanded from four to 12 acres of winter greenhouse pepper production under LED lighting in a $6,458,000 cooperative project with Harrow RDC, Essex Energy and Sollum Technologies. 

Previous generations of commercial greenhouses focused on general, or macro, climate control for whatever was being grown under cover. Operators used manual controls. Computer-operated automatic controls are common today. Now, efforts are moving toward microclimate control using sensor arrays that report to new-generation software.

“Think of the microclimate as a bubble around the plant, a blanket of still air surrounding it. The microclimate can vary across the greenhouse and is strongly influenced by crop structure (geometry) and greenhouse management,” Digweed wrote at the outset of this project. 

Managers learned that uniform control of macroclimate does not guarantee a uniform microclimate. By teasing out the details of climate variables within the canopy, researchers believed they could identify areas prone to disease, better predict yield and even predict the influence of new technology on microclimate before installation. 

In a two-acre zone, sensors won’t pick up the variables that may be happening 300 feet away. More sensors can reduce vulnerability, at a price.

The first objective, and easiest, was to replace high-cost sensors with lower-cost sensors. 

“Normally you’d have a temperature, humidity and CO₂ sensor per zone in your climate control computer for a greenhouse. A commercial scale zone could be two acres. For research, a zone is only 50 square metres,” Digweed says. 

In a two-acre zone, sensors won’t pick up the variables that may be happening 300 feet away. More sensors can reduce vulnerability, at a price. 

“You’re looking at $1,000 for a good, research-grade temperature and humidity sensor. Good quality light sensors are in the range $600 to $1,000 each.”

Collecting detailed microclimate measurements in a commercial greenhouse requires four types of sensors at, ideally, four heights or levels in the crop. That’s 16 sensors for one zone. The new Harrow research greenhouse each have four zones. 

On that very practical cost issue, the Harrow research team began looking for alternate highly accurate sensors at less cost. 

“We found that, if you were careful in what you looked for, you can get a highly accurate sensor at a much lower cost,” he says. 

One type is a class of thermocouple, Type T, normally used for medical applications. “They fit our purposes very nicely. These sensors can be very fine, almost hair-thickness, and aren’t affected strongly by solar radiation or self-heating because they’re so tiny. They can even adhere to the leaf of a pepper plant,” he says.  

A second type of sensor is from the automotive industry. 

“We’ve been able to repurpose some of these sensors to measure temperature and humidity of the air. They are designed for a harsh environment and are generally low cost because they are produced in such high quantities.”

The research also had to come to grips with another sensor issue, communication. Thermocouples and climate computers are traditionally analog; automotive sensors and lower cost data loggers are digital. The most economical array for a given zone combined both types. That issue, too, was resolved in the study. 

“Getting computers to talk to other computers can be challenging,” Digweed says. “Once you’ve got those details sorted out, the results are excellent for accuracy.”

Release of product names and details will depend on permission to publish a peer-reviewed article on the research. Tentatively, that will occur sometime in the second half of 2023.

“We don’t have a publication date yet. Meantime, I’m happy to provide advice to growers and to share what we’ve learned,” Digweed says.

The project began in April 2019 and is scheduled for completion in March 2023. 

Climate modelling progress

Given a few years further development, Digweed suggests, it may be that A.I. systems will eliminate the need for direct microclimate monitoring.

Climate modelling is the second aspect of the AAFC project. The research team is working on a model to predict climate values at heights of one, two, three and four metres based on the actual values at the top height alone. 

Three years later, they have “a robust dataset” for Maureno and Eurix peppers and a mostly complete dataset for tomatoes. The plan to work with cucumbers was set aside but excellent data now is being gathered on eggplant production. 

Based on plant growth data, Digweed has been able to predict photosynthetically active radiation (PAR) values for the peppers at three heights, based on the actual value at a four-metre height. Next, knowing the PAR values, he hopes to be able to predict the effects of temperature, leading to ability to optimize the environment for the plants based on energy input efficiency. 

Important variations in the microclimate have been linked with the type of lighting. This has to be reflected in the eventual software calculations. 

“With the move toward LED lighting, humidity concerns can be an issue,” he says. “In winter, you have overhead lights radiating heat down and heating pipes radiating heat upward, so you end up with a cool area – and moisture buildup –  in the middle of the canopy.” 

The microclimate data is revealing, as well, that a so-called ‘smart’ climate controller program must recognize crop differences. One crop is sensitive, but another isn’t, to winter moisture conditions generated by types of lighting and heating. 

“This isn’t an issue with peppers; they generally have low humidity in winter. But, it is a concern with tomatoes,” he says. “You can see excessive humidity in the middle of the crop canopy in greenhouse tomatoes. It could lead to disease unless accurately predicted and controlled.”    

Emerging technology

The microclimate data is revealing, as well, that a so-called ‘smart’ climate controller program must recognize crop differences.

In recent years, there’s been a big push toward data-drive growing in greenhouses. Growers want to both save energy and improve the climate for the crop. Intelligent, climate-controlling computers are already emerging. 

A follow-up project is underway, funded by the Independent Electricity Systems Operators (IESO) of Ontario and in partnership with Koidra, an artificial intelligence company in Seattle.  

“We’re using our expertise in interior climate monitoring and their expertise in A.I. to try to improve greenhouse cucumber and eggplant production in a commercial greenhouse. It’s a beginning. Growers are aware now of the importance of microclimate, and that we are developing tools for them to optimize it.”

Members of the Harrow AAFC team working with Digweed are physiologist Dr Xiuming Hao, pathologist Dr Genevieve Marchand and Dr Jason Lanoue, post-doctoral fellow. Kenneth Tran, CEO and founder of Koidra, has assigned an internal team of engineers to work with the AAFC group. 

Given a few years further development, Digweed suggests, it may be that A.I. systems will eliminate the need for direct microclimate monitoring. 

Digweed concludes, “If you can predict everything happening in a stable crop based off one or two carefully placed sensors, you eliminate the need for the high density that we built. Low cost would be much more appealing to growers. The system approach we’ve developed is going to be useful for research indefinitely. I could see it accelerating our research programs by reducing the cost for data collection.”