Energy & Operating Cost Savings

Optimizing Desolventizer - Toaster Dome Temperature


January 26, 2015

Industrial plants that are already up and running cannot make drastic changes to improve profits without facing down time. The best path for increasing profit is by implementing small adjustments to existing systems. These optional changes are often done after the basic processes are up and running, and they usually address specific performance and economic goals. Sitting on top of the basic controls, the adjustments are called Advance Process Controls (APC). APC encompasses a broad range of topics, whether a segment of APC (such as advanced regulatory controls) or a type of APC control (such as distributed control system or a supervisory control computer). Every industry that involves continuous process uses some level of APC as a way of limiting expense, controlling production, or maximizing profit and quality within specific constraints. This piece presents an instance of identifying an APC opportunity, designing a solution, and implementing the system in a real-life scenario.

A large soybean processing plant had a Desolventizer-Toaster (DT) located exterior to their facility. Annual temperatures range widely in the region, so the plant allowed dome temperature within their DT to be five degrees higher than necessary as a means of preventing system shut-down if internal temperatures on the DT dropped. They did this because of inherent variability in the process and to prevent loss of revenue due to shut-down. As a result, the plant paid a higher cost than they needed to for the excess steam to maintain the higher temperature. Interstates helped bring the resources together to determine optimal dome temperature within process constraints, to design the control system needed to maintain that temperature, and to implement the system.

Every facility measures and monitors its processes at various steps along the way. As part of the project assessment, process data is collected and analyzed to confirm the business case for the project and set the expectations of the project performance. It's optimal to have twelve months' of data to consult; with this time frame a mathematical model of the system would represent the extremes of summer and winter temperatures, humidity, and other variables. Though twelve months is optimal, we don't always have a full year of data. In this case, we used the historical data that existed, which was approximately six months. Working with our software partner, the data collected was used to construct a mathematical model of this part of the plant's process. Data points (such as temperature, pressure, and flow rates with respect to time) helped identify critical features of each part in the process, how its pieces intertwined, and the degree to which they influenced each other. The highly complex math can be used to predict how changing one part of the process will impact the other parts, how great the degree of impact will be, and even what the ramifications will be for cost and profit. Six months of data included lab data (stored both on paper and electronically), operator log sheets of manual readings, process data, system readouts, and cost-revenue information for raw materials and finished products in current markets. The analysis showed that temperature within the DT dome fluctuated significantly in a six-month period due to process changes. As a result of these process fluctuations, the temperature setpoint was far beyond the desired setpoint to avoid unnecessary shutdowns. Factors influencing dome temperature included steam flow through the DT as well as extractor speed, DT level, and discharge airlock speed.

A good process control model should address contributing factors, but it also accounts for necessary process constraints and limits on how much change the process will bear. Examples of constraints in this case were minimum discharge temperature, maximum dome pressure, and DT electrical load. This particular plant has a DT separate from the Dryer-Cooler (DC), so DT meal discharge moisture percentage was not a limiting factor; for a plant with a stacked DTDC, it likely would have been added to the list of constraints. A rough, two-variable dynamic model of dome temperature was sufficient for our analysis. Using only DT steam flow and extractor speed as our two inputs to the model, we were able to determine that those two variables accounted for 60% of the dome temperature variation the plant was experiencing. This model showed that correctly manipulating steam flow would allow the plant to more accurately control dome temperature.

This particular model's purpose was not to accurately forecast dome temperature to an nth degree, but to tell us whether APC was a cost-effective way for the plant to increase profit. By accounting for 60% of the variability through a single control (steam flow), the model fulfilled its purpose. The analysis estimated that DT dome temperature variation could be reduced by 35-75% if we installed APC for steam flow.

Once there is a working model for the process in question, we are able to use that model to determine return on investment (ROI). This helps a team evaluate whether the suggested changes of controlling steam flow and monitoring temperature are financially beneficial: how much the changes will cost, how much money is expected to be saved over time, and how long before cost is recouped in savings. ROI calculations take into account the engineering and programming costs for the assessment, design, and configuration of the project as well as any necessary instrumentation, hardware and software licensing requirements, support, and updates. It factors in current market prices for inputs (in this case, steam cost) and products (soy meal, oil, etc.). These real-time factors are what give ROI calculations their power: the ability to see immediately how the bottom line will be impacted over time by changing one part of the process. ROI analysis showed that a one degree-Fahrenheit shift in dome temperature corresponded to roughly 583 pph of steam for a constant throughput for a given operating day. Even taking the low end of estimates for cost savings, the math projected close to $70,000 annual savings for restricting the dome temperature by only 2.5 degrees. This savings is gained when control is only 50% greater than they already had! This estimate also does not include savings associated with condensate loss reductions due to decreased steam usage or additional uptime due to no unplanned outages. The team quickly concluded the plant would benefit from adding a new steam control valve, a new temperature transmitter on the DT dome, and the APC to oversee the two.

There are two options for this sort of APC: a simple control solution with modular multi-variable control (MMC) blocks, or more complex integrated quality/through-put control with a multi-variable model predictive control (MPC). MPC is an APC that usually resides on a supervisory control computer the process already uses for basic functioning. MPC identifies important process variables and the dynamic relationships between them. It has been a prominent part of APC ever since computers acquired the necessary computational capabilities to perform matrix-math based control and optimization algorithms in the 1980s. This computing ability allows MPC to control multiple variables simultaneously with complete understanding of their complex relationships in real time. Because of the number of independent and dependent variables and the complexity of the relationships concerning dome temperature, the team recommended a design solution using an MPC controller. The controller will use manipulated variables such as steam flow to influence the controlled variable of dome temperature. It will also consider controlled variables such as minimum discharge temperature, DT motor amps, and maximum dome pressure to keep its steam flow manipulations within necessary constraints. Using disturbance variables that influence the controlled variables, the MPC controller knows how much the discharge temp will drop before it changes steam flow, thereby altering the steam flow without under correcting or over correcting.

For this situation, we want the MPC to use these

Control Variables:

  • Control dome temperature to a target value
  • Maintain discharge temperature above a limit
  • Maintain dome pressure within limits
  • Maintain gearbox amperage within set limits
  • Maintain deck pressure within set limits

Disturbance Variables:

  • DT Top Deck Level
  • Extractor Speed

Manipulated Variable:

  • Steam Control Valve

All of these goals and restrictions were written in the MPC programming at the time of implementation. Cost for the hardware (MPC controller, SCADA servers, operator HMIs, PLC controllers and Ethernet capability), software programming, installation, and on-going support were calculated during ROI, and installation and implementation went smoothly.

The recommended MPC control began testing and monitoring in August 2014. Our data collection from August through October show a marked decrease in steam consumption and 50% decrease in dome temperature variance compared to the data before APC installation. Initial ROI numbers forecast that savings for the plant would surpass cost in nineteen months. The real-time data from daily operations indicate the break-even point to be less than twelve months. Savings on the project is equal to approximately 1 pound of steam per bushel of soybeans processed. Since the implementation of the project, the plant has experienced no unplanned shutdowns due to the dome temperature dropping below its safe threshold. Necessary constraints have been honored with no instances of operation that violate any of the MPC disturbance variables, just as we hoped. The plant is on track to save more than $150,000 annually as a result of installing one APC to control one variable.

For continuous industrial processes already in production, improving operational efficiency or increasing productivity can come from implementing some form of APC on top of basic processes. This project was a small step to see how APC might help control a widely-variant DT dome temperature. Beyond controlling steam flow for this single process, the plant now has a path forward to greater profit with small risk. This same process flow has close to a dozen other points that may benefit from further APC control, leading to even greater cost savings within the plant. Other plants within the company could also improve their profit margins by implementing the APC measures used successfully at this facility. The company could improve profit margin greatly over the long-term without hiring personnel, without increasing consumption, without purchasing more raw materials, and without changing product pricing. APC provides a logistically sound way to manage resources better within specified boundaries. This specific APC implementation is a good example of how seemingly small adjustments can provide substantial results over time.