Software process analyzers provide more frequent, robust and often more cost-effective measurements than are available from traditional laboratory information systems. This online software based analyzer allows an operator or process engineer to monitor the real-time feedback and predictions of future process performance, and make manual adjustments to the process to ensure desired end quality.
This same model-based solution can also be deployed online to provide real-time control and optimization. The APC solution continuously monitors a number of key process variables minute-by-minute, predicts outcomes based on the established model of the process, feeds the real-time and predicted data back into the model, simulates the impact on the plant’s objectives and then issues commands back to the Distributed Control System (DCS) to make changes to the key variables that are driving results. This solution addresses the second tenet outlined by the PAT Guidance framework, “continuously reducing the variability caused by the process.” APC improves management and the optimization of a plant operation in four main areas:
- Enhanced ability to determine product quality in real-time, on-site without waiting for a lab sample or an end-of-the-run analysis.
- This capability is provided by the ability to predict quality accurately in-process and the improved process stability.
- Improved operations stability and consistency.
- The product and batch transitions are performed more quickly and consistently. This capability translates directly to improved operational efficiencies and transition losses are reduced.
- Ability to push process constraints.
- APC reduces process variability, enabling manufacturers to run closer to desired targets and push to constraints with more confidence. The result is more throughput, higher quality product or both.
- Increased operational efficiency.
- This is measured in terms of reduced raw materials and energy consumption.
A food industry success story
Fonterra Co-operative Group Ltd.’s dairy ingredients group, NZMP, began to implement APC at its facilities in 1995, to improve processing efficiency for milk proteins, powders, cheese and casein. The company realized an ROI of over 60%, and also saw its production rates improve by 5 to 15%, its product quality by 50%, energy efficiency by 5 to 12%, while product yields increased and variability fell, for all key control variables.
As a result of the performance gains, the company has installed APC at many of its 23 New Zealand-based operations and Fonterra’s majority-controlled overseas joint ventures.
Let us consider one specific implementation, a milk powder plant making nutritional, whole milk and skim milk powders. The plant has two evaporators and one spray dryer with the two falling-film evaporators feeding a Niro atomizing spray dryer processing 6.5 to 9 tons per hour.
The evaporators concentrate the milk to approximately 50% total solids (TS), with each evaporator using two effects. There is Mechanical Vapor Recompression (MVR) on the first effect and Thermal Vapor Recompression (TVR) on the second effect. After passing through the evaporators, the concentrate is dried in the dryer chamber and then in the static fluid bed (SFB) and the two vibrating fluid beds (VFB’s). The final moisture specification limit is typically set between 2.5 and 3.5%.
This APC project has been implemented in two phases. In the first phase, Fonterra’s on-line grading analysis system was installed to allow In-Process Testing (IPT) results to be measured and reported statistically. This allows operators to view product composition results relative to product specifications and budget goals. The on-line grading analysis system also provides a means to interface IPT results to control applications.
In the second phase, a Pavilion APC application was developed to minimize process variability by compensating for disturbances, with the intention of creating a performance-driven, “obedient” plant. Multivariable predictive control software was used to implement the evaporator and dryer application.
In addition, as part of the implementation, a software-based analyzer was deployed to predict moisture for continuous feedback to the control application. A wealth of historical data for dryer’s wide product mix allowed a model to be created that reflected moisture ranges for a variety of products with a high degree of accuracy. The analyzer, a Pavilion Soft Sensor, was enhanced by installing an application that biased the prediction hourly with IPT data from the online grading analysis system.
Control applications were developed for each evaporator feeding the dryer with the goal of reducing total solids (TS) variability by 50%. An audit conducted in April 2003 demonstrated that the APC solution had exceeded this objective, reducing the variation by approximately 73% for Evaporator 1 and 68% for Evaporator 2. The reduction in standard deviation of evaporator total solids will allow operations to increase total solids targets in the evaporators without violating viscosity limits. This allows increased water removal in the more thermally efficient evaporators. A higher concentrate total solids additionally allows increased dryer throughput. A 0.5% increase in total solids from the evaporators could lead to a 2% increase in dryer throughput.
Figure 2 and Figure 3 (below) show the final total solids from Evaporator 1, before and after APC control. Note that variation is noticeably improved when the APC controller is running.