AptiTune
Download AptiTune white paper (PDF)
Download AptiTune brochure (PDF)
Have a look at the AptiTune Demo (Movie)
Introduction
AptiTune™ is a multivariable PID tuning software package designed to provide a comprehensive PID tuning solution for complex heat integrated process units, where interactions between the various PID loops are significant.
AptiTune™ delivers several unique benefits:
- Easily tunes PID loops that ‘fight’ each other
- Delivers results in just a few minutes
- Often achieves ‘one shot’ tuning
- Easy to use yet provides reliable and safe results
- Significantly improving process productivity
Key Features
The new AptiTune™ multivariable PID tuning software package has been designed to provide a comprehensive PID tuning solution for complex heat integrated process units where interactions between the various PID loops are significant. The software is specifically designed to provide the following capabilities:
- Ensure a mild learning curve: The software was designed to be very easy to learn, and even easier to use. The high quality Graphical User Interface makes it very obvious what user decisions are required next, yet it provides sophisticated analysis tools to help the user understand the impact of the controller design decisions.
- An easy and reliable way to tune PID loops that “fight” each other: For the first time, control engineers can simultaneously optimize the tuning of two or more PID loops that interact in an undesirable way, and make the loops coexist in a peaceful way.
- Productivity Improvements: The AptiTune™ software executes rapidly and new tuning can be derived in just a few minutes, once the required model has been provided. In addition, the tuning can be entered on the DCS exactly as calculated, usually with no need for further iterative improvement.
- PID tuning that is robust with respect to model mismatch: Models of moderate accuracy can be used, which helps to reduce the time spent step testing the process, increasing the productivity of the control engineers. In addition, the tuning can withstand significant changes in the dynamics of the process as the unit moves to different operating points, without going unstable.
- The software is well integrated in the standard APC workflow:AptiTune™ starts with an FIR model file in mdl format. The model file represents the OP to PV responses for the entire nxn model matrix (n=nr of loops), for the process scope under investigation. The model is typically derived from a limited amount of either open loop or a closed loop step test data.
- AptiTune™ can optimize very large control systems:No explicit limits are imposed on the maximum number of PID loops that can be tuned at the same time. The software has been tested on a 20 loop example, and the solution converged in about 10 minutes, while 3x3 or smaller problems complete in less than 30 seconds. Single loop examples only take 1-2 seconds. The execution speed and available RAM of the host computer will determine the total state space model order that can be represented. In practice, the availability of test data will determine how many PID loops can be optimized simultaneously.
- Intuitive workflow: The control engineer supplies the controller output (OP) and instrument (PV) ranges, then select the DCS specific PID equations to use, and finally choose the type of controller (either P, PI, PID, or I-only). The possibly noisy Finite Impulse Response (FIR) model is then smoothed by first applying a central average filter to the FIR coefficients, and then fitting a high order state space model with explicit dead time, in order to the approximate the FIR model in a realistic way. The state space model is usually very smooth with clean dead time, and will display a high degree of process realism. This approach ensures that we get reliable tuning, without having to rely on over-simplistic first order plus dead time models, even if the test data set was noisy or relatively short.
- Calculates process specific tuning: It is straight forward to achieve closed loop process behavior that is precisely tailored to the specific requirements of the particular process. The software does not rely on over-simplistic tuning rules, but rather makes use of an accurate high order state space model of the process, DCS specific PID equations, and a sophisticated non-linear optimizer to achieve the best possible trade-off between competing design objectives like setpoint tracking, disturbance rejection and stability robustness. The user has full control over how to balance these trade-offs.
- Defaults specifications that actually work: The control engineer would normally use the default specifications to start with. The default values almost always provide tuning that works well straight away. If the engineer has specific process objectives in mind, like safety requirements or specific performance objectives, the user can inspect the simulated OP and PV responses for both SP steps and load disturbances, and modify the trade-off weights in order to make certain loops faster or slower. For example, for a furnace coil outlet temperature controller, we would normally want the controller output which cascades to the fuel gas pressure controller not to overshoot the steady state value by more than 10%. Explicit specifications can be provided in the time domain by limiting the maximum OP value for a specified SP step. All hard constraints are specified in the same units as the process variables.
- Explicitly specify the closed loop PV rise time per loop: The user can limit PV overshoot if needed, and put limits on the peak OP value for a specified step change in the SP. The engineer can make one loop faster at the expense of another loop, or improve the load disturbance rejection at the expense of the setpoint tracking response, etc. We can also constrain how much noise in the process variable (PV) will be amplified by the controller, by explicitly specifying the maximum acceptable noise level in the OP for a specific amount of noise in the instrument signal.
- Productivity improvements: What is most impressive about the software is that it often delivers “one-shot tuning”: The engineer iterates in the safe environment of the host computer (a high end personal computer), where the optimizer can try our several thousand different values for the PID tuning in a matter of seconds. As long as we have used a reasonably accurate OP to PV model, we will almost always be able to type in the new calculated tuning on all of the interactive PID loops on the DCS system at the same time, and it will generally work as simulated. If the model was good enough, we can then make an SP step on the DCS system, and we will observe almost identical performance on the DCS screens compared to what we saw on the simulation graphs on the PC. There is normally no need to try out multiple tuning sets on the DCS, often saving us days or weeks of engineering effort, with less risk to the process, while significantly improving process performance.
- Insurance against changes in process dynamics: We have all been there: We tuned a PID loop manually and achieved really fast performance with good damping, only for the loop to start cycling really badly 2 weeks later when the feed rate to the process unit changed. AptiTune™ solves this problem convincingly. The optimizer uses a sophisticated robust control approach and explicitly takes expected worst case changes in process gain and dead time into account in order to ensure that all of the interacting control loops will be stable and well damped even if the process dynamics change quite dramatically in future. This feature also provides protection against model identification errors, and ensures that a moderately accurate OP to PV model is entirely adequate for PID tuning purposes.
License Options for Aptitune
have a look at the possible license options .
Presentations and Demos on Aptitune
Have a look at the AptiTune Demo (Movie)
New features in AptiTune V2 (Movie)
Aptitune Wizard for integrating processes (Movie)
Aptitune Wizard for non-integrating processes (Movie)
