An oil extraction plant is a continuous production, where every extra tenth of a percent of oil in the meal or deviation in the miscel is directly converted into financial losses. It would seem that with such sensitivity to quality indicators, oil content control should be as operational as possible. However, most enterprises still work according to the scheme: they took a sample, sent it to the laboratory and received the result in 2–4 hours. During this time, the line has already processed 50–200 tons of raw materials. The question is not whether technological solutions for online control exist. They exist and have long been tested in the West. The question is why they have not become a standard and what really prevents their implementation.
NIR analyzers vs. laboratory - what is the fundamental difference?
Near-infrared spectroscopy (NIR) allows you to measure the oil, moisture, protein and fiber content in a material stream in real time - without sampling, without reagents, without the participation of a laboratory technician. The sensor is installed directly on the conveyor, in the auger or in the mixer pipeline, and generates data every 5–30 seconds.
A classic laboratory is a multi-stage process: sample collection, transportation, preparation, extraction using the Soxhlet method or similar, weighing, and recording the result in a journal. Even with ideal organization, it takes at least 1.5–2 hours from the moment of collection to the number in the journal. In practice, it takes 3–4 hours, taking into account the queue and staff workload.
The main difference is not in accuracy. Modern NIR systems provide an error of ±0.1–0.2% in oil content, which is quite comparable to the laboratory method with correct calibration. The main difference is in the time and number of measurements. The laboratory gives 6–12 data points per day. The NIR analyzer — thousands.
What does NIR measure in an oil extraction plant:
- Residual oil in the meal after extraction is a key indicator of the efficiency of the process.
- The moisture content of the meal at the outlet is critical for safe storage and high-quality pelleting.
- The concentration of oil in the mist is the basis for controlling the operation of the distillation unit.
- Oil content and moisture content of seeds at the input - input control for adjusting technological parameters.
The fundamental advantage of NIR is that it is not a replacement for the laboratory, but rather a complement to it. The laboratory remains the reference and is used for calibration and verification. However, operational process control is made possible by online data, not in its absence.
Data Delay = Lost Money
This is not a metaphor. This is arithmetic.
With a line capacity of 500 t/day and a laboratory cycle of 3 hours, the plant “blindly” processes about 62 tons of sunflower between two control points. If at this time the process “went” towards an increased residual oil content of the meal by 0.3% above the norm, this is a direct shortage of oil. At a price of $ 1,100 / t, losses for one such episode amount to $ 60–80. For a year, with several such deviations per week, the figure grows to $ 150,000–300,000.
Let's consider three specific scenarios.
Scenario 1. Increased residual oil in the meal. Violation of the extraction regime — solvent temperature or seed level in the extractor — leads to an increase in residual oil content from 0.7% to 1.1%. During laboratory control, the operator learns about this after 2–3 hours. All this time, the plant actually “sells” the oil together with the meal at the price of the meal.
Scenario 2. The oil content of the input raw material has changed. A new batch of seeds has arrived from another region — the oil content is lower by 2.5 pp. Without online input control, the oil yield is recalculated post facto. The planned change indicators have already been calculated “according to the old norm” — and the technologist does not know whether to look for a problem in the process or accept the fact of a change in the quality of the raw material.
Scenario 3. Meal moisture beyond the norm. Meal with a moisture content of more than 10–11% when packaged is a risk of spontaneous combustion and spoilage. If laboratory analysis shows an excess after the batch has been stored or loaded into wagons, there will be either a claim from the buyer or costs for re-drying and reloading.
What is common in all three scenarios is that the decision is made after the deviation, not during it. Online measurement takes the operator from “firefighter” mode to “pilot” mode — he sees the instruments and reacts to the disaster.
How to "live" without online data and how much it costs
Most domestic plants compensate for the lack of online control with a system of “protective buffers”: stricter internal standards for residual oil content, increased sampling frequency, additional staff in the laboratory, manual checks by the operator “by touch and color.” This system works — but it is expensive.
Direct costs to compensate for the lack of online control:
- Intentionally understated standards — the plant undershoots oil output in order to have a "safety margin" without operational data.
- Increased costs for reagents and equipment due to the increased number of laboratory tests.
- The salary of night lab technicians means the process doesn't stop, which means the control doesn't stop either.
- Costs for complaints and reprocessing of batches with deviations in meal moisture content.
- Penalties for quality indicators when delivering for processing or export.
There are also indirect costs that are more difficult to calculate. The lack of operational data means making decisions based on the experience and intuition of the technologist, rather than on actual process indicators. This makes it impossible to automate accounting at oil extraction plants in the full sense: without a continuous flow of high-quality data, no MES system or accounting platform can build a reliable material balance in real time. As a result, the report on oil output is generated at the end of the shift or day - and is only a snapshot of the past, not a tool for managing the present.
It is worth mentioning the personnel risk separately. A plant that “lives” without online data critically depends on specific people — an experienced technologist, an attentive laboratory assistant, a responsible operator. As soon as this person goes on vacation, is fired or gets sick — the quality of control drops. The NIR system does not get sick or go on vacation.
Why online monitoring is still not the standard
There are several reasons, and none of them are technical.
First, the high initial cost. A full-fledged NIR system for an OEZ with several measurement points costs 80,000–200,000 EUR, depending on the configuration and manufacturer. For a company with an old investment culture, where it is customary to “repair, not replace”, this is a psychologically difficult threshold.
Secondly, the difficulty of justifying ROI within the company. Losses from data delay are distributed over time, invisible at the level of a single shift, and not reflected as a separate line in the P&L. The director sees the costs of the NIR system — and does not see the “line” of money saved, because it is “dissolved” in overall efficiency.
Third, distrust of new measurement methods. "We've been testing oil content using the Soxhlet method for 30 years - why do we need a spectrometer?" is a real position that can be heard at production meetings. Overcoming this barrier requires not only a technical but also a managerial solution.
Fourth, there is no market pressure. As long as competitors operate in the same way, there is no competitive incentive to invest in more precise control. The situation changes when entering more demanding markets or when players appear who have already invested in online analytics and have a lower cost of product.
