Busy bee concept for feed production process control

The largest Norwegian feed compounder Felleskjøpet Agri, has as the first feed company in the world installed a real time, instant feedback NIR (Near Infra Red) system in one of their factories. The system, called APIS Spektron Feed, was installed in autumn 2007 and has been thoroughly tested. APIS is also an acronym for Advanced Plant Information System. The system measures feed quality parameters and instantly presents the results as numbers and graphs. The well known NIR technology is combined with specially designed software to achieve a system that gives instant feedback to operators, supervisors and quality personnel, enabling fast corrective actions. Manual feed analysis results (e.g. fines or durability) as well as relevant process parameters (e.g. temperature) may also be included and presented.
“The philosophy behind Spektron Feed is to gather, process and present data of importance for feed producers. NIR measurements give little meaning for operators unless they are presented together with target values and limits.
Similarly, production in tonne per hour is presented together with the maximum capacity for that particular feed and graphically displayed if the target is reached,” explains Håkon Standal, project manager at Prediktor. “Other data, like temperature measurements, could be recorded and presented only in reports – only when temperatures are outside defined values alarms are shown for the operator. To achieve this, Spektron Feed is communicating with equipment and software present in the factory.”
To process data Spektron Feed utilises real time software and has been developed together with Felleskjøpet Agri. New software is constantly added to the system. Graphs presenting actual product cost are one of these features. A module for energy monitoring is also under development.
As Standal already indicated there are significant differences between Spektron Feed and traditional NIR systems used in laboratories. The Spektron measurement probe (or gauge) is a self operating NIR-unit with optics, spectrometer and a computer. The probe may be remotely monitored to assure good quality measurements. The probe is designed specifically for industrial applications and can be placed in a rough environment. Placed in conveyors or over conveyor belts it gives real time monitoring of protein, fat and water in feed ingredients or finished feeds. Features include:

• Presentation of recipe refined targets and tolerances (min/max values);
• Measurements and presentation updates may be done every second;
• Immediate alarms when measurements are outside defined tolerances;
• Continuous presentation of actual product cost;
• Reports with report filters, graphs and numbers;
• Web based system – easy access over company intranet or Internet for monitoring and maintenance.

Cost control
An interesting feature of the system is a real time overview of the costs of production and the product. The cost control shows the real cost of the finished feed – including also the “loss” when a feed with cheaper ingredients is run through the pellet press, but at a speed that utilises only half of the capacity. Product cost is calculated with the use of real time software, enabling results to be continuously presented in graphs/trends on the operator screen, using ingredient cost – sent from recipe optimisation software (Format or similar) and is automatically updated in the APIS Spektron Feed system. Other inputs are the cost of running the line (cost/hour), which is preset for each line by the feed miller. Furthermore the system monitors the actual production (tonne/hour), which is collected continually from the factory control system.

Spektron Feed in practice
The first two systems were placed at Felleskjøpet Agri’s feed mill in Stange in 2007. The system paid for itself in less than a year with savings on ingredients cost. “We have chosen the Spektron system, because they were able to supply the inline NIR probe as well as the software,” says Vebjørn Nilsen, Process Manager at Felleskjøpet Agri. “Control of raw materials is very important, because there is a lot of variation. Protein content in barley, for example, can vary from 8.5 to 12%,” he adds. Last year Felleskjøpet Agri started a new project with Prediktor to continually measure grain quality and this year the cooperative intends to install more systems in other feed mills and expand the grain quality project.

Felleskjøpet Agri supplies guaranteed salmonella free feeds to its farmers. To establish this, the feed mills are equipped with heat treatment systems of Bühler and Kahl. Of course these systems limit maximum throughput in the mills. “That is why we have put strict control measures in our mills,” explains Nilsen.
“Labour costs are high and we try to automate as much as possible. That is also one of the reasons we installed the inline NIR-sampling. With this system we can in real time analyse the feed making process from raw materials to final product.”
Nilsen expects much from the real time production analysis of the system. “We have determined our costs for production, such as electricity and steam, maintenance, labour and interest costs for equipment and put these data into Spektron Feed.
As a result we can view the costs per tonne while producing. If, for example, the operator increases throughput in the mill, we see the costs per tonne go down. And if, for example, our nutritionists develop a feed with cheaper ingredients, it might well be that the costs of producing this feed might be higher. So we can take immediate corrective action, such as adding processing aids to increase production per hour, change ingredient composition to reduce troublesome ingredients or modify the recipe,” Nilsen says.
“The cost control is a feature one would typically find in a Manufacturing Execution System (MES) or Process Development Execution System (PDES) and really is a tool for improvement of production,” Standal says. “All results and measurements can be found in reports. We can also make this cost control into a closed loop. That is, make models that suggest a corrective action for the operator, or even change machine settings or add processing aids automatically. But this could of course only be done with a high degree of automation in the factory,” Standal adds.