Optimise your
batch processes
Predictive adaptive control applied to material transfers in batch processes

Batch processes (batching, blending, and filling) depend heavily on the speed and repeatability with which each material transfer (ingredient addition) is completed for every recipe executed. Each and every transfer requires precise cut-off control over the valves, screw feeders or pumps, as those transfers directly impact the annual profitability of a manufacturing facility. Creating a cost-effective material transfer control system that consistently improves process quality and throughput while reducing raw material waste and operating costs might seem to be an impossible task. Mettler Toledo has patented a technology to do just that.

While the process industry has developed model-based adaptive/predictive control strategies for continuous processes, up until now there has been no commercially available model-based material delivery system for batch processes. Realising the large potential material savings that would be enabled by optimisation of material transfers, one of the world’s largest consumer goods companies invested a significant amount of effort into studying the use of predictive adaptive control for material transfers, as well as determining the root causes of inaccuracy and material waste at higher transfer speeds. The result was the PAC (Predictive Adaptive Control) technology for material transfers (ingredient addition) which, over the last ten years, the company successfully deployed and benefited from throughout its manufacturing facilities. Mettler Toledo has exclusively licensed this technology and expanded upon its capabilities. In addition, the company has studied and accumulated the application expertise to compliment the PAC technology.

How PAC technology works
The PAC technology generates improvements by building a real-time mathematical model of the material transfer process. This enables the material transfer control system to predict, during the current transfer, the exact moment material cut-off needs to occur to attain the optimum result. The PAC technology automatically compensates for normal process variants that cause errors, eliminating costly raw material over- and underfeeds. Because cut-off is determined in real time, there is no need to slow the material feed down
(as is done by most systems) to gain control over the variability in the process. This also eliminates the need for multispeed feeders or extra gates and valves. The
material transfer can be executed using “Simple”, “Fast”, “Singlespeed”, “On-Off” control. Batch cycle time is reduced, enabling higher manufacturing throughput. The need for costly and complex multi-speed feed control systems is eliminated, reducing design engineering, maintenance and capital equipment costs for control valves, pumps and piping. The PAC technology is self-tuning. It “learns” the characteristics of a delivery system, and then adapts to the normal variations that occur over time.
The PAC control technology and additional material transfer functionality have been integrated into Mettler Toledo’s field-proven “Jagxtreme” terminal product line. This new Material Transfer Controller, as this category of controller is called, is known as the Q.iMPACT matroller. A matroller is a combination of measurement hardware and material transfer control technology. The Q.iMPACT matroller is a key component in the Q.i suite of products. Integrating Q.iMPACT matrollers as complimentary optimisers of batch control systems provides performance enhancing opportunities for existing and new batch process control systems. Integration is currently available with Rockwell Automation batch control systems, Honeywell batch control systems and, in the near future, with Siemens batch control systems.
Improve control with PAC
The material transfer study that was completed by the consumer goods manufacturer in their early research of the
PAC technology revealed four specific sources that contribute to losses in
quality and subsequent material waste
in batch processes. With this knowledge built into the PAC technology, these
factors can be controlled and significant levels of improvement in material delivery gained. The four error-producing sources are:
- Instrument “reading” lags which occur in all measurement systems due to analogue-to-digital conversion delays. These delays introduce time slewed measurement errors. It is especially important to consider these errors just prior to the point of material flow cut-off.
- Material-in-suspension at the point of cut-off is always present. Its variability needs to be considered and compensated for in determining cut-off.
- Material that passes through the valves after the close command has been given (valve let-through) must compensate for determining cut-off.
- The kinetic energy stored in the material, which is released as it impacts other material in the mixing vessel, causes “phantom” weight readings and can impact material cut-off.
All of the above errors are dynamic. They have their foundations in the normal, constantly changing, processes found in any batch control system during a material transfer. They are also additive, and therefore, influence the repeatability of material transfers. While some of these error sources can be reduced by good design practice (e.g. material-in-suspension), others are much more difficult to control as they vary constantly and unpredictably over time. The aggregate of these errors is referred to in this document as “spill”. Spill is a combination of measurement lag, unmeasured material, and kinetic energy. The challenge is to know how much spill will occur prior to executing a cut-off. The spill must then be used to compensate for process change by introducing a positive or negative off-set (adjustment) to the previously requested material transfer target weight (setpoint) as requested by the recipe or formulation card. Use of this control technique results in a more accurate material transfer.
The premise of the PAC technology is simple. If you can accurately predict a dynamically changing spill during the feed stage of the material transfer and quickly adapt for it just prior to cut-off, you can attain significant improvement in material transfer repeatability and product quality. That translates directly into cost savings. If you can achieve this level of dynamic control, there is no need to constantly slow the process down at an end of each feed and loose valuable manufacturing time. This level of control, enabled by the PAC technology, is embedded in the Q.iMPACT matrollers. The PAC technology has two components that help achieve the goal of obtaining optimum material transfer cutoff. The first is the predictive component and the second is the adaptive component.
Predictive Control: The predictive component of the PAC technology automatically calculates, in real-time, instrument reading lags, material-in-suspension, valve let-through, and material kinetic forces. It then uses this data to predict the exact moment of cut-off for optimum accuracy in the current feed. It offers three sets of PAC algorithms to meet your specific needs:
- Spill only algorithm – used to control material movements that have very low flow rates.
- K1 algorithm used to control material movements that have moderate flow rates.
- K2 algorithm used to control high flow rates.
Adaptive Tuning: The adaptive component of PAC technology automatically tunes the algorithms to compensate for normal variations or changes that occur in all processes including:
- material density,
- flow rate,
- valve characteristics (speed/let-through),
- pump performance,
- and other variability.
This adaptive tuning eliminates the need for periodic maintenance or manual tuning of the PAC algorithms. It therefore has a positive impact on cost of ownership, as it reduces the need for the setpoint trimming done in most systems today.
Reduce buffer setpoints and save
raw materials
Simply improving the ability to deliver materials more accurately at a constant high speed may reduce process variation, but it does not necessarily save raw materials. Over- and under-fills typically balance each other out regardless of the control. Therefore, nearly the same amount of material would still be used in both cases if the setpoint remained the same. However, in current multiple-speed control systems, setpoints are usually “cushioned, fattened or buffered” (terms used by operators) and are increased to help guarantee minimum material transfers are achieved. This practice is instituted by operators to achieve desired tolerances even on poor feeds, and is used to limit the need for reworking batches.
Improved control capability, on the other hand, can greatly reduce the need for buffering. With the improved capability realised with the use of PAC technology, we can take advantage of tighter achievable, tolerances, and modify and reduce the target/setpoint. Using the improved control, we can guarantee the feed tolerance at a lower setpoint and reap significant material savings. Some PAC installations have been financed totally by the material savings generated by this type performance improvement.
Simplify control methods
and save on capital equipment
With the PAC algorithms integrated into Q.iMPACT matroller a simpler, more cost-effective, robust material transfer control system can be achieved. PAC technology can be applied for the delivery of materials, using Load-Cell Gain-In-Weight, Load-Cell Loss-In-Weight and Flow Meter addition techniques. The technology has been extensively and successfully deployed in high volume batch manufacturing since the early 1990s. All of the applications used a combination of flow metering and gain in weight/loss in weight measurement for material transfer control. Flow meters and scales may be used independently of each other or in combination. The PAC technology uses simple, cost-effective single-speed On/Off control, and still achieves higher control repeatability than could be realised with more expensive two speed-feeds at comparable delivery times.
Simplify design and reduce
custom automation
Designing an automated material delivery operation requires a significant amount of work, and can consume a large part of any process controller if that is where the control is to be accomplished. However, by migrating the Feed Start, Feeding, Feed Stop and Feed Finish stages (4) out of the controller and into the Q.iMPACT matroller, we are able to remove a significant amount of the custom automation that is normally required, as well as the need to continually support it. Studies indicate that through use of the Q.iMPACT matroller at least 60 to 80% of the custom programming normally required in a controller-based material delivery system is eliminated by moving control of the material transfer process into the matroller. Again, this relieves the control engineer of having to design and manage custom code. In addition, with the PAC technology embedded in the matroller, the control of the material transfer process will be located closer to the actual point of measurement where it can be more responsive and deterministic.
Improving batch cycle time
and improve asset utilisation
On/Off control improves batch cycle time by eliminating the traditional multi-speed control typically used to control cut-off and to slow down the material transfer process. Multi-speed flow control extends batch cycle time by introducing an additional material transfer step referred to as “slow feed”. Slow feed allows the process to “creep”, so cut-off is slowed down to accommodate measurement and process lags. PAC technology eliminates multi-speed feed slow step, delivering significantly better repeatability, shorter batch cycle time and improved manufacturing throughput. It also eliminates the previous constraint of an engineering tradeoff between repeatability and speed.
Using PAC technology, material delivery times have been reduced by 20 to 30%, which has translated in overall batch cycle time improvements of 7 to 10%. This specifically talks to improved asset utilisation by increasing the number batches an existing batch manufacturing system can produce and possibly deferring larger capital investments in additional batch manufacturing systems.j