How to improve CNC machining processes
In machining processes such as milling, predicting manufacturing time and surface quality is not an easy task, especially in the manufacture of products with more complex geometries. Despite the significant technological advance that has occurred in recent years such as the Internet of Things (IoT), cyber-physical media, cloud computing or big data, as well as new production techniques, such as metal additive manufacturing, material theft is still the main industrial process today. These innovations present technological challenges yet to be resolved., such as milling complex shapes. This means that the control of the machines must be able to read many lines of code to be able to reach the programmed advance at all times and not slow down.. But even the most advanced CNC machines are not capable of reading data that fast.. This leads to a number of problems, such as surface quality, tool malfunctions and, above all, generates an uncertainty in the machining time.
Current CAM systems estimate the machining time considering that the machine will move according to the programmed forward speed. Therefore, the inaccuracy of this estimate can reach errors greater than 1000%, which makes the industry not use the time estimation provided by this software. This limitation hinders the entire production schedule of an industry, making integration with production support systems impossible, like MRP and ERP. A recent study investigates this technological limitation and presents alternatives to predict the real time of milling complex shapes on CNC machines.
The latest advances in manufacturing processes in the metallurgical sector, like additive manufacturing, have allowed print metal products with mechanical properties very similar to those manufactured by conventional methods, eliminating many limitations in the manufacture of complex geometries or allowing the manufacture of internal cavities. However, the process of subtractive fabrication of material still occupies an important space in the production chain, either for finishing and adjusting tolerances, or even for low cost manufacturing, where fits the production of parts composed of primitive geometries such as cylinders and hexagons.
Turning and milling machining process
Between manufacturing processes with subtraction of material, the turning and milling are more important due to their volume of application. Among them, milling is the most used to manufacture pieces with more complex shapes, which is applied in the industries of molds and dies, aeronautics and energy.
Although this process is well known, predict manufacturing time and the quality of the surface is not an easy task, especially in the manufacture of products with more complex geometries where the finishing process is done with ball tools and the contact between the part and the tool is constantly changing during the process. Besides, the finishing process occupies an important part of the manufacturing time and is directly related to the quality of the final product. As an example, in the mold and die industry, where a mirror finish and better surface quality are required, and then significantly reduce polishing time.
In the production process of any product, you have to plan the manufacturing time, since it directly affects the Delivery time, cost and production cycles. Each process can have a greater or lesser effect. In small batch production, accurate estimation is essential when budgeting, and if the product has significant dimensions, the effects on production time can even compromise the production and profits of the company. While, in large batch production, experimental validation can be amortized by transferring the cost to all products. In this case, is most important in the choice phase of the production process.
From the current point of view, companies are tending to implementation of virtual twins that faithfully represent the production processes. For it, it is necessary to correctly model both the machines and their behaviors, thus correctly representing their productive capacities.
The linear interpolation is the method generally used to represent a toolpath on a complex surface, described by small line segments, using only G01 functions, according to DIN standard 66025. Los cartesian points represent the beginning and end of the straight segment. The length of the straight segments used in the CNC program is related to the calculation tolerance, defined by the user in the CAM software, associated with the curvature of the surface.
In the trajectory calculation process, CAM software calculates cutter contact point (CC) and then make it a cut location (CL). Using the following equation:
In this process, these points that describe the trajectory must respect a maximum error user supported, for this a tolerance range is used during the calculation process.
Although it is capable of generating similar toolpaths, los algorithms used in the CAM software to calculate the toolpaths are developed by different software vendors. This way, algorithms can be different. In a precise analysis of such algorithms, It can be observed that different software calculates the 'same' toolpath. However, looking in detail, there are important differences in the trajectories generated by the different programs.
The studies presented have shown that the calculation algorithms used by the CAM software can influence the whole process, diverging between systems: time to calculate trajectory, time to post-process the NC program, size of the generated NC programs and disparity in the generated points, directly affecting the quality of the machined surface and the process time due to fluctuations in the feed rate.
The machines have different processing speeds of the CNC program, either due to mechanical or processing limitations, that affect the speed of production. The calculation of the response time of the machine (Machine Response Time – MRT), is defined as the time that the machine takes to process and execute the movement of a line segment.
Numerical control performance in machining
Los more advanced numerical controls allow working with extensive CNC programs (2GB), removing programming en bloc and its drawbacks. However, there are some factors that still influence the performance of the CNC and, consequently, limit the machining feed rate:
- block processing time (TPB): is the average time required for the numerical control process a program line and send information for the activation of the servomotors. The length of the straight segment used to describe a part of the toolpath (linear interpolation of straight segments), together with the TPB, are factors that limit machining progress.
- Clock y Bus: the frequency of the 'clock' must be related to the number of CPUs in the system and the capacity of the bus. A CPU of 32 bits will have poor performance if you use a bus 16 bits.
- Block Buffer: the CNC stores already processed blocks in a temporary area ('lung' – Block Buffer). This way, there is always a processed CNC block waiting to be executed by the server. Conventional CNCs are generally capable of store up 10 blocks processed. To work at high cutting speeds – HSC (High Speed Cutting), more blocks are required to be processed in the Block Buffer.
When the programmed feedrate is greater than the capacity of the CNC to interpret and send information from movement to servo motors, there are limitations in the advance that can cause two different situations, depending on the characteristics of the numerical control used:
- outdated controls, with resource limitations when the programmed speed exceeds the capacity of the CNC. For this reason, are produced bumps in the machine during the machining process, causing discontinuities in movement and resulting in an unsatisfactory finish.
- modern controls. If the same situation occurs will slow down forward until it snaps to a value you can manage. With this, a proper surface finish is achieved, but the speed of progress of the process is reduced.
Response time of a CNC machining machine
When machining complex geometric shapes, the feed value can fluctuate considerably during machining. These oscillations are due to technological and dynamic limitations of the machine and the numerical control equipment (CNC). Each team has its characteristics, with large performance differences between current machines.
To identify the performance of a machine with a given CNC and identify movement limitations for machining complex shapes, the methodology called TRM (Machine Response Time). The method to obtain this variable is presented below:
- A CNC program is developed that contains linear movements of straight lines, using the code G01, with small increments, in the direction of the axis to be analyzed. The path must travel a path significantly greater than the distance of one increment at a programmed high speed, greater than 5.000 mm/min. Due to technological limitations it is known that even using modern and high speed CNC machines, they are not yet capable of executing this movement at high forward speeds.
- the cnc program must be run on the specific machine. During the movement of the machine, the operator must observe on the CNC screen the feed rate reached along the path – Fmax.
- The valor Fmax will be used in the equation 2 to get the TRM.
With the variation of the segment size it is possible to obtain the response profile of the machine and predict its behavior according to the tolerance applied in the generation of the program..
This methodology used has already proven to be quite efficient at predicting the timing of programs that use 3 axes. This work contributes to the study of multi-axis machines that seek to understand the relationship of linear and angular axes.. Thus, the machining time forecast is investigated with 4 y 5 axes where the forecast error may be greater than the 1000%.
Experimental procedure of mechanized machines of 5 axes
To evaluate the calculation errors predicted by the CNC machine control assembly, two machines have been evaluated 5 axes with rotary axes A and C. When using freeform geometries that require movement of at least 4 axes for finishing, errors relative to actual machining time are expected to be larger.
Posteriorly, the time predicted by the CAM program (with a tolerance of one tenth and one hundredth) was compared with that calculated by the CNC simulator and the real machining time of the machine. during machining, was used the 100% of spindle rotation and the empty movement, simulating machining with the 100% in advance.
During the experimental process, all trajectory smoothing features disabled, to verify the real movement value of the set of numerical controls of the machines. Besides, 'tool-tip' function not used, that does not allow kinematic inversion in multi-axis programs.
After verifying that the magnitude of the forecast errors, both in the CAM software and in the internal simulator of the machine, They were big, the methodology that was developed for linear motion (3 axes), was applied to try reduce errors related to machining time estimation.
The same methodology of linear axes was used for the angular axes.. In the analysis, it was discovered that for angular kinematics the size of the formed trace is irrelevant for processing, although they can be related to each other, the limitation is in the reduction of the angular increase.
The linear and rotary axes were first evaluated independently. To obtain the initial parameters of the segment of the experimental matrix, programs were created that reduce the size of the segment (lineal o angular) until a reduction in advance is found.
After individual tests of each rotary axis, combinations of axes were made and it was found that for the machines evaluated:
- Each linear axis has control of independent movement, with the longest route length always determining the longest segment time.
- Los angular axes are slower that the linear axes and the first have dominance by mixing the response time of the segments with the mixed movement.
With the individual tests, a methodology has been carried out in which the choice between the time of each follow-up considering angular and linear movement has been proposed.. The methodology was applied to programs of 4 axes for the 2 geometries.
How to improve machining times
All individually tested axes of the CNC machines showed results with a linear response pattern.. If the mechanical limits of the machine axes are not exceeded, the Control throttling is the main performance limiting component during the milling process.
It seems that the set of CNC machines evaluated have different response times.. In both cases, the pattern has an exponential character and reducing the segment size exponentially affects the execution time of the program.
It is known that the tolerance affects surface quality y, if possible, keeping it as high as possible within its specifications can yield significant gains for large-scale production.
To improve the budget it is necessary to carry out a processing in the CNC program in order to reduce the estimate in machining time. For it, the following methodology is exposed:
- Identify linear and angular motions of each line of the CNC program (referring to a linear or arc increment)
- The choice of follow-up time is always by the smallest increment, since it is the limit.
- Since angular motion is slower than linear motion, if it is used as a reference to calculate the arc up to a certain value, where the reduction in the size of the line segment becomes dominant. are required Choice conditions for each machine in order to compare whether the linear or angular domain is used in the calculation according to the increments.
The tests carried out showed that the simultaneous movement of a linear axis plus an angular axis has the slowest time. This way, fast axis needs to slow down to maintain trajectory. The response of the angular axes is lower than that of the two linear axes and it is necessary to calculate the average time of the segment in milling of 4 y 5 axes.
It was also observed that the processing time for each axis of the evaluated machines is independent., so the total time for each segment is limited by the smallest increment.