GA Engineering Inc. is a Canadian-incorporated engineering SME based in British Columbia. Alongside our delivery work, we are exploring how modern data tooling, machine-learning techniques, and AI-assisted methods can improve the efficiency of hot mix asphalt plants โ particularly around burner control, mix recipe queue planning, and energy use.
The work described below is an early-stage research direction, not a packaged offering. We are at the stage of scoping problems with operators, framing data requirements, and prototyping. We're sharing it openly because we'd like to talk to plant operators, OEMs, researchers, and funding programs interested in the same questions.
Each of these is grounded in a specific control problem on a hot mix asphalt plant. None of them assume a fully autonomous plant โ the operator stays central in the loop.
Drum-burner setpoints today are tuned by operators based on experience and feel. We're exploring data-driven setpoint suggestions that take into account aggregate moisture, ambient conditions, RAP percentage, and target rate โ with the aim of reducing fuel per ton without giving up mix quality. Models would run alongside the existing burner controller as an advisory layer, not as a direct replacement.
Plants typically run several mixes over a paving day. The order and timing of recipe changes affects silo utilization, asphalt-binder usage, and downtime between mixes. We're prototyping queue-planning tools that look at the day's job list, silo levels, and material availability, and propose a sequence that reduces mix-change loss.
Fuel use per ton, electrical demand, and baghouse pressure trends carry information about overall plant health. We're working on models that turn historian data into shift- and job-level energy summaries, with anomaly flags that point operators at likely causes before they become alarms.
Most plants already collect more data than they use. We're exploring lightweight predictive-maintenance signals โ bearing temperature drift, motor current trends and hours of operation, baghouse ฮ-P patterns โ derived from the data already on the plant historian, without adding new sensors.
A few principles we are committed to in this work, ahead of any specific model or tool:
A pragmatic mix of industrial and ML tooling:
GA Engineering Inc. is a Canadian-incorporated small enterprise based in Vancouver, British Columbia. The principal is a Professional Engineer registered with Engineers and Geoscientists of British Columbia (EGBC). Our service work pays the bills today; the R&D direction described on this page is what we believe will keep Canadian asphalt plant automation competitive in the next decades.
If you are a producer, OEM, researcher, or program officer focused on advancing industrial AI/ML applications within heavy industry, we would welcome the opportunity to connect and discuss potential synergies.
Producers willing to share anonymized historian data in exchange for early access to tools and findings. Even a single plant's full season of data is useful at this stage.
Burner, controls, and plant OEMs who would like to see an open advisory layer alongside their existing automation platforms, rather than against them.
Academic groups, applied research centres, and innovation programs interested in industrial AI / ML applied to a heavy-industry use case.
Whether you're a producer, an OEM, a researcher, or a government program officer โ we'd be glad to talk.
Get in touch