FunctionBay Technical Support Site | RecurDyn Solver, the Present and the Future

RecurDyn Solver, the Present and the Future

Ho-Young Cha, Team Lead

Solver Team 2, FunctionBay, Inc.

▶ Would you introduce yourself first?

▷ Hello, everyone. I am Ho-Young Cha, Team Lead for Solver Team 2 of Development Group at FunctionBay. I joined FunctionBay in 2008 right after graduate school. It has been 10 years already. I feel like I still have a very young spirit, but I was very surprised while I was writing my introduction because I realized how fast the time has passed. Actually, since I graduated from the graduate school, joined the company, and got married all in 2008, this year which is the 10th year anniversary is personally very meaningful to me, and it became even more meaningful as I got this opportunity to introduce myself through the company newsletter.

▶ How long have you participated in the development of RecurDyn and what was the motivation to join FunctionBay?

▷ First, I will tell you how I joined FunctionBay.

I was in charge of developing Fuzzy Block of CoLink under the guidance of Professor Sung-Soo Lim of Kyunghee University when I was in graduate school for my master’s degree. At that time, the person in charge of the project was Deok-Jin Yoon, currently the Chief Manger who is researching the next generation of Solver for RecurDyn on the Basic R&D Team at FunctionBay. This project opened the door for me to directly visit FunctionBay and provided me with an opportunity to experience the atmosphere of the company. Since I was very interested in programming, I hoped to join FunctionBay, and the project led to an opportunity to formally join and work for FunctionBay immediately after graduation.

▶ What was the first version you directly participated in developing and could you give us an example comparing the version at that time and the current version?

▷ The first version I developed since joining the company in 2008 was V7R2. As for Solver, the V7R2 version was released not long after the introduction of 64-bit Solver product.

I studied control engineering for my master’s degree, and in the beginning, I participated in the development related to eigenvalue analysis with CoLink. Not long after joining the company, I became in charge of developing RFlex, which is a flexible body similar to eigenvalue analysis.

▶ What was the most robust function in that version, and how did that function evolve and get adopted in the current version of RecurDyn?

▷ It is my personal opinion, but I believe most people would agree. It is the analysis of MFBD which includes the flexible body. There are two flexible bodies that are supported by RecurDyn, which are FFlex and RFlex. Although only limited analysis was possible with a product from a competitor, in RecurDyn, it was called FFlex, which is short for Full Flex, meaning that “complete analysis of a flexible body” is possible. This is still recognized as a differentiated function of RecurDyn even today, in 2018. At that time, an external Mesher had to be used in order to use the FFlex flexible body, but now, it has been improved so that it is possible to create a flexible body using only RecurDyn. Today, it has evolved even further, and it has advanced to a function that can conduct analysis by freely changing to a rigid body and a flexible body (not only RFlex but also FFlex) through the technology called G-Modeling. Since it is favored by customers as it can be easily changed to a rigid body and a flexible body, there are many requests for the additional development related to it.

▶ If you have to pick one area that RecurDyn Solver really stands out compared to products from competitors, what would it be?

▷ First of all, the biggest strength compared to competitors’ products is that analysis of MFBD including Rigid, RFlex and FFlex is possible. And I am confident that RecurDyn is the most competitive in terms of the analysis of Nonlinear Transient/Dynamics.

Secondly, I would pick multidisciplinary integrated analysis based on MFBD (Multi-Flexible Body Dynamics) as another strength of RecurDyn. For example, RecurDyn allows easy analysis with Block Diagram modeling which is typically used in the control part called CoLink. The analysis of a flexible body with not only CoLink but also other CAE software that supports standard interface called FMI is possible. Using FMI offers the advantage of including analysis results of various areas that can’t be analyzed by RecurDyn such as AMESim, MATLAB, and SimulationX. In addition, our own standard interface called SPI (Standard Particle Interface) was developed for the last version, V9R1. Based on this, flexibility analysis with the software based on a particle method became possible. Furthermore, the interface with Particleworks and EDEM has been developed and is currently being used. I expect that more interfaces for a greater variety of particle method software would be developed. I would like to say that the area that RecurDyn can be utilized would expand more and more as time passes as the integrated analysis with the noise-vibration area of the gear (NVH, Noise & Vibration & Harshness) would become possible in future.

▶ As RecurDyn’s version continues to be updated, Solver gets updated as well. What is the area you are focusing the most on regarding the enhancement of Solver?

▷ Surely, what we focus on as the highest priority in Solver is the fast analysis speed with accuracy as prerequisite. Then, the next priority is stability and integrity of the analysis results.

All development and enhancement issues of Solver usually start from the requirements of customers who use RecurDyn and the correction of the bugs that have been reported. And it is being enhanced based on the actual analysis models that are gathered from various customers. I believe that enhancement of Solver based on the actual models used for actual tasks rather than the models used in the development environment is the growth engine of RecurDyn.

▶ What do you think is the biggest change as the version has changed from V8 to V9?

▷ First of all, we found that the effect of numerical damping of the integrator is quite big in the versions leading up to V8, and therefore, it was greatly improved in V9. In other words, numerical damping parameter of the integrator has to be reduced in order to reduce the numerical damping of the integrator until V8. But in V9, an accurate result can be acquired without separately revising the parameter as the effect of the numerical damping has been minimized.

Secondly, we are aiming to make an expansion regarding the multidisciplinary integrated analysis based on MFBD. For future versions, we should be able to provide a solution against the noise and vibration of gear system.

And if I can say one more thing from the perspective of the developer of Solver, I would like to point out that we have changed the development environment of Solver to be the most advanced. The development environment is like a tool that the developer uses, and it has a benefit of improving the working environment. In particular, we have brought about a direct positive effect on the accuracy and speed by changing the compiler and updating the library to the latest version as well as the benefit of improving the work efficiency in various areas.

Advanced Hybrid Integrator that is to be included as the default integrator starting with RecurDyn V9R2

▶ What are the details and background story of the integrator that would be newly applied to V9R2? What is the advantage that the user would get by using this?

▷ In V9R1 which is the previous version, we had made an improvement on the numerical damping having a big impact on the Implicit G-Alpha integrator. But Implicit G-Alpha integrator can only be applied to the analysis that includes rigid body and RFlex.

As such, in V9R2, we will further expand the improvement we had made to include FFlex. We will also additionally offer “Advanced Hybrid Integrator”. In other words, the solution is being maintained by keeping the previous Hybrid Integrator, and “Advanced Hybrid Integrator” has been added as a new solution.

In summary, three functions have been applied to V9R2 at large. First, it has been enhanced not to be affected by numerical damping even for a problem that includes FFlex. Second, the analysisn speed including RFlex has been enhanced even further. Third, the integrator has been improved so that it can be applied not only to Rigid and RFlex but also to FFlex. In the previous version, I would especially like to explain further about the damping effect; with regard to the gear, vibration and noise, the high speed rotation and excitation of various frequency ranges must be considered. If the damping effect is significant, the resonance effect that is most important for the noise and vibration does not appear much. The new integrator offers an advantage of raising the accuracy of the analysis of the noise and vibration including FFlex.

Lastly, the new integrator has enhanced analysis speed for the problem that includes RFlex. If I put it more specifically, there is a greater degree of enhancement with regard to the analysis speed when the RFlex body uses over 200 modes and there are more entities (joints, forces) that are connected to RFlex.

▶ Would you like to say something to the readers and users before wrapping up?

▷ Although developing RecurDyn is my daily task, I realized that I have grown very much as I wrote down and summarized the work I have done to date. I hope customers feel a big improvement when they use RecurDyn for their work. I will wrap up by saying that I will put even more effort from now on so that I can provide the best solution as the developer of software that is leading the CAE market.