Archives: Cases

MPC is used for control of polymerization temperature and reaction rates, as well as final product quality, by controlling cooling and heating utilities and metering (dosing) rates of initiators and monomers.

The main benefits of using this technology compared to conventional control are:

• Accurate control of reactor temperature and reaction rates.
• Increased productivity by maximization of monomer feed rates, within constraints determined by available cooling capacity, pressure constraints, and other quality and safety related constraints.
• Control of final product quality in accordance with specifications.

The essential features of the CENIT technology, which enables the above benefits, are:

• Online parameter estimation – the model adapts online to changes in process conditions.
• Predictive control – the future process performance is continuously predicted and optimized.

Process modelling

The mechanistic process model utilized within Cybernetica CENIT is adapted to each specific process unit. Reactor design data are acquired from the customer. Other model parameters are estimated off-line from logged process data and lab measurements using Cybernetica ModelFit.

Model of semi-batch emulsion co-polymerization reactor

• Reaction kinetics for the specific system of hydrophilic and hydrophobic monomers.
• Phase equilibria and phase transfer rates.
• Energy balances for reactor with cooling and heating utilities.
• Quality parameters (e.g., Mn, Mw) are estimated from moments of the MWD and validated from analytical measurements.

Model predictive control (MPC)

The MPC controller provides the following functionality:

1. Accurate temperature control by steam heating and jacket cooling.
2. Metering (dosing) of monomers and initiator: Maximization of monomer feed rates, within constraints determined by available cooling capacity, pressure constraints, and other quality and safety related constraints.
3. Control of polymer quality: The temperature profile is optimized such that the specified average molecular weight (Mn, Mw) is reached and the polymerization time is minimized.

Safe operation

The CENIT system will ensure that the polymerization reactions are safely controlled in spite of uncertain model parameters. The adaptive model will compensate for deviations between the predicted and the actual reaction heat, as well as varying cooling capacity throughout the batch.

In addition to the inherent safety associated with the MPC methodology and the online model adaptation, the CENIT system includes fault detection and diagnosis functionality.

Implementation

Cybernetica CENIT applications typically run on a dedicated application station (Windows server). It communicates with a DCS system via the Open Platform Communications (OPC) protocol.

Cybernetica employees have extensive experience in industrial implementation and commissioning of advanced process control (APC) applications.

Integration with sequential batch control system

Cybernetica CENIT can run independently of a batch control system, with the operator providing essential inputs such as recipe specific parameters. CENIT can, however, preferably be integrated with a sequential batch control system, in which case the batch control system will provide a number of input parameters to CENIT, dependent on the current stage in the batch process and on the specific polymer grade being manufactured.

Maintenance

Cybernetica’s maintenance programme ensures that all applications perform at their best at all times. Cybernetica rapidly responds to customer requests. Cybernetica CENIT has built-in functionality for reproduction and diagnosis of current or past process situations, facilitating efficient analyses of issues reported by the customer.

The most common and important issues are reported directly to the operators, enabling the operators to act when necessary.

The system optimizes temperature profile and dosing of initiators, with the purpose of minimizing polymerization time while meeting target quality specifications and fulfilling a number of process constraints. MPC is used for accurate temperature control in accordance with the optimized profile, by controlling heating and cooling utilities, and process water temperature.

The main benefits of using this technology compared to conventional control are:

• Increased productivity through optimization of temperature profile and initiator dosing, and accurate control in accordance with the optimized profile. The cooling capacity can usually be utilized close to its maximum during the entire polymerization phase.
• Accurate control of quality parameters (e.g., “K-number”) in accordance with target specifications.
• Condenser cooling capacity is maximized by controlling the venting of inert gases from the condenser, with minimum loss of VCM gas through the venting system.
• The MPC detects any drop in cooling capacity and, if necessary, determines the required amount of inhibitor before any significant temperature deviation has occurred.

Process modelling

The mechanistic process model utilized within Cybernetica CENIT is adapted to each specific process unit. Reactor design data are acquired from the customer. Other model parameters are estimated off-line from logged process data using Cybernetica ModelFit.

Model of suspension PVC polymerization autoclave

• Energy balances for suspension, cooling jacket and reflux condenser.
• Reaction kinetics:
  • Free radical polymerization mechanism
• Thermodynamic calculations:
  • Phase equilibria: Distribution of monomer between the four phases.
  • Reactor pressure.
• Quality model:
  • Quality parameters (e.g., Mn, Mw) are estimated from moments of the MWD and correlated to viscosity (“K-number”).

Estimation of model parameter profiles

Accurate prediction of cooling capacity and reaction heat is ensured as follows:

• Heat transfer coefficients for cooling utilities are estimated as conversion dependent parameter profiles, to account for the changes in these parameters during the progress of a single batch. The estimated parameter profiles are also continuously updated to account for drift in heat transfer capabilities over a longer time scale.
• Any deviation between the reaction heat predicted by the kinetic model and the reaction heat observed from process measurements are compensated for by the estimation of conversion dependent corrections to the kinetic model.

Run-to-run optimization

The temperature profile and the quantities of initiators are optimized at the start of each batch:

• The optimization is based on an economic criterion, weighting cost of batch time versus cost of each available initiator.
• The optimization is performed such that the target product quality and a number of safety related and other process constraints are fulfilled.

Model predictive control (MPC)

MPC enables accurate control in accordance with the optimized batch profile:

• The initial polymerization temperature is reached fast and energy efficient by controlling the temperature of process water during the reactor charging phase.
• The MPC controls the flow rates of cooling water to the jacket and condenser cooling systems, and distributes the available cooling water between the cooling utilities in an optimal way.
• Venting of inert gases from the condenser is controlled based on the profile of the estimated condenser heat transfer coefficient, with the purpose of maximizing cooling capacity while minimizing loss of VCM gas through the venting system.
• The MPC predicts if the heat of reaction will exceed the available cooling capacity, and determines the necessary amount of inhibitor to regain control of the polymerization temperature. This might occur due to an unexpected drop in cooling capacity, e.g. due to foaming.

Safe operation

The batch optimization will ensure that the predicted reaction heat does not exceed the calculated cooling capacity profile. Still, the cooling capacity might change during the batch due to changes in inlet cooling water temperature or other unexpected changes. However, the MPC controller will detect early in case the cooling capacity is not sufficient to control the temperature in accordance with the temperature reference profile, and the operators will be advised to charge a small amount of inhibitor in this situation, before the loss of cooling capacity has resulted in any significant deviation from the temperature reference trajectory.

In addition to the inherent safety associated with the MPC methodology and the online model adaptation, the CENIT system includes fault detection and diagnosis functionality.

Implementation

Cybernetica CENIT applications typically run on a dedicated application station (Windows server). It communicates with a DCS system via the Open Platform Communications (OPC) protocol.

Cybernetica employees have extensive experience in industrial implementation and commissioning of advanced process control (APC) applications.

Integration with sequential batch control system

Cybernetica CENIT can run independently of a batch control system, with the operator providing essential inputs. CENIT can, however, preferably be integrated with a sequential batch control system, in which case the batch control system will provide a number of input parameters to CENIT, dependent on the current stage in the batch process and on the specific PVC grade being manufactured.

Maintenance

Cybernetica’s maintenance programme ensures that all applications perform at their best at all times. Cybernetica rapidly responds to customer requests. Cybernetica CENIT has built-in functionality for reproduction and diagnosis of current or past process situations, facilitating efficient analyses of issues reported by the customer.

The most common and important issues are reported directly to the operators, enabling the operators to act when necessary.

MPC is used for control of reactor temperature, pH and reaction rates of Urea-Formaldehyde (UF) and Melamine-Urea-Formaldehyde (MUF) manufacturing processes, by controlling cooling and heating utilities and dosing of base and acid.

The main benefits of using this technology compared to conventional control are:

• Accurate control of pH during dosing of base and acid. Amounts of base and acid are minimized to avoid overbuffering of the resin.
• Consistent reaction rates during condensation, using both inline pH and viscometer, and by controlling the viscosity rate of change.
• Accurate temperature control.
• Shorter and less variations in batch cycle times.
• Improved safety by preventing condensation runaway and gelling.
• Accurate endpoint determination (cut-off).
• Less need to take grab samples from the reactor.
• Less operator intervention – higher degree of automation.

The essential features of the CENIT technology, which enables the above benefits, are:

• Online parameter estimation – the model adapts online to changes in process conditions.
• Predictive control – the future process performance is continuously predicted and optimized.

Process modelling

The mechanistic process model utilized within Cybernetica CENIT is adapted to each specific process unit. Reactor design data are acquired from the customer. Other model parameters are estimated off-line from logged process data using Cybernetica ModelFit.

Model of UF/MUF resin reactor

• Kinetics and thermodynamics:
  • Methylolation reactions
  • Condensation reactions
• pH model
• Viscosity advancement model
• Cooling and heating utilities:
  • Coil cooling is dependent on estimated heat transfer coefficients and on the liquid level in the reactor.
  • Reflux condenser and vacuum system.
  • Steam heating.
• Feed systems for raw materials

Viscosity advancement model

The viscosity advancement is predicted based on the kinetic model. Since the viscosity depends on many factors, it must be updated from viscometer measurements.

Model predictive control (MPC)

MPC enables safe and accurate control of pH, temperature and viscosity advancement by controlling:

• Dosing of base and acid.
• Coil cooling, jacket cooling and steam heating.
• Vacuum and condenser cooling during polymerization and distillation.

The viscosity rate of change model is used to keep the viscosity advancement rate within maximum and minimum limits during condensation, and to adjust the pH by dosing base or acid as required.

Safe operation

The use of both inline pH and viscometer measurements, in combination with the pH and viscosity advancement models, are essential to reduce the risk of runaway reactions and gelling of the reactor content during condensation. pH probes might drift over time and viscometers might fail to give accurate viscosity readings. By applying both measurements and by requiring both the pH and the viscosity advancement to be controlled within safe limits, reaction runway due to single sensor faults is avoided.

In addition to the inherent safety associated with the MPC methodology and the online model adaptation, the CENIT system includes fault detection and diagnosis functionality.

Operator graphic for polymerization monitoring and endpoint prediction

The system includes an operator graphic display where the profiles of temperature, reaction heat versus cooling capacity, pH, viscosity and viscosity rate of change are compared to those of a previously manufactured “golden batch”.

The purpose of the system is to:

• Monitor the development of the current batch, as compared to the golden batch.
• Detect deviations between the current batch and the golden batch.
• Predict the future progress of the current batch, including the endpoints.

A snapshot of the graphic display during condensation is shown in the figure below. The history is displayed to the left of the vertical line indicating present time. The graphic display is used to monitor the current batch (solid lines) by comparing with a previously manufactured “golden batch” (dashed lines). The viscosity and the target viscosity are shown in the upper graph; the viscosity rate of change and the minimum and maximum limits are shown in the middle graph; the pH and the minimum and maximum limits are shown in the lower graph.

Implementation

Cybernetica CENIT applications typically run on a dedicated application station (Windows server). It communicates with a DCS system via the Open Platform Communications (OPC) protocol.

Cybernetica employees have extensive experience in industrial implementation and commissioning of advanced process control (APC) applications.

Integration with sequential batch control system

Cybernetica CENIT can run independently of a batch control system, with the operator providing essential inputs such as temperature setpoint profiles and raw material amounts. CENIT can, however, preferably be integrated with a sequential batch control system, in which case the batch control system will provide a number of input parameters to CENIT, dependent on the current stage in the batch process and on the specific resin grade being manufactured.

Maintenance

Cybernetica’s maintenance programme ensures that all applications perform at their best at all times. Cybernetica rapidly responds to customer requests. Cybernetica CENIT has built-in functionality for reproduction and diagnosis of current or past process situations, facilitating efficient analyses of issues reported by the customer.

The most common and important issues are reported directly to the operators, enabling the operators to act when necessary.

Optimization by control of impurities

By specifying the product impurities, plant personnel may optimize the production from different perspectives:

• High setpoint values, close to the customer specified maximum levels, decreases the energy consumption and thus the need of steam to the reboilers. This implies energy optimization through “minimum energy consumption” for the specific distillation column
• For NGL products with significant price difference, the separation should be done in such a way that the flow of the most valuable product is maximized. This implies higher impurity for the high value component and lower impurity for the low value component. This strategy represents a simplified kind of economic optimization, not considering the energy costs.
• Only average values (on storage tank) are critical, thus higher impurities may be allowed (outside spec) for shorter periods in order to increase capacity during transients (e.g. temporary feed increase)

Optionally, a detailed economic optimizer, handling feed compositions, product prices, flow routing and energy consumption, could be implemented on the top of the different MPC applications. The different MPCs will thus constitute a required control layer in order to achieve the economical optimum.

Requirements for the process control to handle feed changes and allow for flexible flow routing between trains of columns, have made implementation of advanced control critical in order to operate the plant without increasing the control room resources.

Implementation

The implementation of a number of MPC applications at Statoil’s gas processing facilities has been carried out through a tight cooperation between personnel in Statoil and Cybernetica.  Cybernetica has been involved because of the company’s experience in the development and implementation of MPC. The applications are implemented in “SEPTIC” – Statoil’s in-house tool for MPC. A total number of approximately 25 MPC applications were implemented at two different sites, and Cybernetica-personnel were involved in all of them. The activities lasted for about 3-5 years, including initial application maintenance and hand-over to local personnel. A detailed study of the different PID controller loops was performed ahead of the implementations, covering re-tuning of more than 100 PID-controllers.

Application details

None of the basic PI(D) controller loops in the plant are removed. The MPC applications therefore regards the setpoints of the basic controllers as it’s free or manipulated variables. For a pure two-point distillation column, this typically means

• the septoint of reflux flow controller (FIC)
• the setpoint of column bottom temperature controller (TIC)

Setpoints for level controllers and pressure controllers are used as manipulated variables in a few cases. If not used, the values are typically given from design requirements.

The main disturbances of the separation units are the changes in feed flow and feed composition. These variables are not manipulated, but the MPC application detects the variations and compensate for the effects by changing the manipulated variables.

Dynamic models have therefore been established from each of the input variables, both disturbances and manipulated variables, to each of the controlled variables. The controlled variables are typically

• Light (top) product impurity, that means concentration of heavy component
• Heavy (bottom) product impurity, that means concentration of light component
• Differential pressure across the column

The impurities are typically estimated by soft sensors, based on column temperatures and pressure, and updated from gas chromatographs when new sample analysis are available. The differential pressure is used as an indication of column load, and will typically only be controlled below a certain maximum value, in order to prevent column flooding.00

MPC is used for control of reactor temperature and reaction rates, by controlling cooling and heating utilities and metering (dosing) rates of formaldehyde, phenol and catalysts.

The main benefits of using this technology compared to conventional control are:

• Stable, accurate and consistent control of reactor temperature and reaction rates.
• Shorter and less variations in batch cycle times.
• Improved consistency of product quality parameters.
• Improved safety.
• Endpoint (cut-off) determination using kinetic model.
• Less need to take grab samples from the reactor.
• Less operator intervention – higher degree of automation.

The essential features of the Cybernetica CENIT, which enable these benefits, are:

• Online parameter estimation – the model adapts online to changes in process conditions.
• Predictive control – the future process performance is continuously predicted and optimized.

Process modelling

The mechanistic process model utilized within Cybernetica CENIT is adapted to each specific process unit. Reactor design data are acquired from the customer. Other model parameters are estimated off-line from logged process data using Cybernetica ModelFit.

Model of phenolic resin reactor

• Kinetics and thermodynamics:
  • Methylolation reactions
  • Condensation reactions
  • Formaldehyde, hemiformal and ionization equilibria
• Cooling and heating utilities:
  • Coil cooling is dependent on estimated heat transfer coefficients and on the liquid level in the reactor.
  • Reflux condenser and vacuum system.
  • Steam heating.
• Feed systems for raw materials

Model predictive control (MPC)

MPC enables safe and accurate control of reaction rate and reactor temperature by controlling:

• Coil cooling, jacket cooling and steam heating.
• Vacuum and condenser cooling during polymerization and distillation.
• Metering (dosing) rates of formaldehyde, phenol and catalysts.

MPC allows fast response to step changes in temperature setpoints, which will reduce batch cycle times. Predictive temperature control is illustrated in the video below.

 

Temperature control during initial exotherm and step change of temperature setpoint. The predicted cooling water flow and reactor temperature, during the 15 minutes prediction horizon, are shown to the right of the vertical line indicating present time. A 15 minutes history is shown to the left.

Safe and efficient metering of raw materials

The metering (dosing) rates of raw materials are safely controlled such that the predicted reaction heat does not exceed the currently available cooling capacity, and such that other constraints (e.g., maximum reactor temperature rate of change) are not violated. The available cooling capacity may typically vary significantly during a batch due to

• changes in cooling water temperatures,
• changes in available cooling water flow rates,
• varying heat transfer coefficients,
• and varying liquid level in the reactor.

All these changes are immediately compensated for in the calculation of raw material metering rates.

Optimal metering rates of formaldehyde will typically vary greatly from start to end of the metering phase, and there is a lot to gain in batch cycle time, in addition to the safer operation, by utilizing MPC.

Safe operation

Cybernetica CENIT ensures that the polymerization reactions are safely controlled in spite of uncertain and rapidly changing cooling capacity. The adaptive model will also compensate for deviations between the reaction heat predicted by the model and the actual reaction heat observed from reactor measurements.

In addition to the inherent safety associated with the model predictive control methodology and the online model adaptation, CENIT includes functionality for fault detection and diagnosis.

Automatic endpoint determination

Endpoint determination of “shortly condensed” resins of low viscosity development are often accomplished by manually analysing grab samples from the reactor, e.g. for free phenol content or based on cloud point determination. This is undesirable when full automation is aimed for.

The degree of polymerization, as calculated from the kinetic model, is successfully being used for endpoint determination of shortly condensed resins, resulting in improved consistency of product quality parameters.

Operator graphic for polymerization monitoring

The system includes an operator graphic display where temperature, reaction heat and cooling capacity profiles are compared to those of a previously manufactured “golden batch”. The “golden batch” profile is “time warped” to make the two batches comparable even though the reaction rates progress differently with time. The purpose of the polymerization monitoring system is to:

• Monitor the development of the current batch, as compared to the golden batch.
• Detect deviations between the current batch and the golden batch.
• Predict the future progress of the current batch, including the endpoints.

A snapshot of the graphic display is shown in the figure below. The history is displayed to the left of the vertical line indicating present time, and the future predictions are shown to the right. The graphic display is used to monitor the current batch (solid lines) by comparing with a previously manufactured “golden batch” (dashed lines). The reaction heat and the maximum cooling capacity are shown in the upper graph; the reactor temperature and the temperature setpoint are shown in the middle graph. A selection of user defined variables can be shown in the lower graph.

 

Implementation

Cybernetica CENIT applications typically runs on a dedicated application station (Windows server). It communicates with a DCS system via the Open Platform Communications (OPC) protocol.

Cybernetica employees have extensive experience in industrial implementation and commissioning of advanced process control (APC) applications.

Integration with sequential batch control system

Cybernetica CENIT can run independently of a batch control system, with the operator providing essential inputs such as temperature setpoint profiles and raw material amounts. CENIT can, however, preferably be integrated with a sequential batch control system, in which case the batch control system will provide a number of input parameters to CENIT, dependent on the current stage in the batch process and on the specific resin grade being manufactured.

Maintenance

Cybernetica’s maintenance programme ensures that all applications perform at their best at all times. Cybernetica rapidly responds to customer requests. Cybernetica CENIT has built-in functionality for reproduction and diagnosis of current or past process situations, facilitating efficient analyses of issues reported by the customer.

The most common and important issues are reported directly to the operators, enabling the operators to act when necessary.