Challenges and Lessons Learned for the Design and Implementation of Large PROFIBUS Network

Size: px

Start display at page:

Download "Challenges and Lessons Learned for the Design and Implementation of Large PROFIBUS Network"

Kellie Black
6 years ago
Views:

1 ABSTRACT for the 215 ISA WWAC Symposium Challenges and Lessons Learned for the Design and Implementation of Large PROFIBUS Network Francisco Alcalá, PE 1 * and James Powell, P.Eng. 2 1 CDM Smith, 231 Maitland Center Parkway, Suite 3 Maitland, Florida, USA, (* alcalaf@cdmsmith.com and Phone: ) 2 Siemens Canada Limited, 1954 Technology Drive Peterborough ON K9J 6X7, Canada SUBMISSION TYPE 6-12 page paper plus 3-minute presentation KEYWORDS PROFIBUS, Network, Profibus DP, Profibus PA, Intelligent Plant Instrumentation, Communication layers, network components, Smart network design, Installation, Large network, Design Constraint ABSTRACT The onset of new and emerging technologies such as Intelligent Plant Instrumentation (IPI) is bringing fresh challenges to the design of instrumentation systems such as an interesting level of complexity in the size and design of the network. For example, large networks need to be designed to meet the necessary process requirements, and at the same time maintain the advantages that the smart instrumentation is able to provide through emerging digital data management systems. The design of data networks for IPI implies a 18 degree change in the vision of the system when compared to a typical hardwired system. The evaluation of the layers of network communication and their constraints becomes an important factor that needs to be addressed when designing large systems. OBJECTIVE The objective of the paper is to describe the requirements, constrains, and best practices of the design and construction of a large plant PROFIBUS network. A set of milestones will be presented in order to provide a model that could be used in similar projects. PROFIBUS DEFINITION AND TYPES PROFIBUS, which is short for process fieldbus, is an industrial control network introduced in It can use many different physical layers such as RS-485, Fiber optics, and Manchester Encoded Bus Powered (MBP) which is defined in IEC standard. PROFIBUS also incorporates a number of different profile standards to make device integration easier. The PROFIBUS standard has been released three times; DPV, DPV1, DPV2. All versions are interoperable due to the fact that the packet design was well defined in DPV and then never modified. Devices written to the different versions can all co-exist on the same network with the only limitation being in the controller. If the controller is written to DPV, then it can go into data exchange with all of its slaves, but it cannot use DPV1 or DPV2 functionality with any of its DPV1 or DPV2 slaves since it does not know how. Aug 4-6, 215 Orlando, Florida, USA

Alcalá, Powell 2 DPV defined the basic protocol while DPV1 introduced functions required for process control call PROFIBUS PA (Process Automation) such as MBP physical layer which can be IS, and the

2 Alcalá, Powell 2 DPV defined the basic protocol while DPV1 introduced functions required for process control call PROFIBUS PA (Process Automation) such as MBP physical layer which can be IS, and the ability to configure devices over the bus. DPV1 also introduced PROFIsafe which allowed PROFIBUS with emergency shutdown systems. DPV2 added functions required for PROFIdrive which is used for high speed drives applications. For the application presented here, only DPV and DPV1 functionalitity is required How PROFIBUS Works: Figure 1 PROFIBUS standards PROFIBUS is a straight forward master-slave type protocol with the addition of a token ring on top to coordinate multi-masters. The protocol defines two types of masters: a class 1 master as a traditional controller such a PLC or DCS, and a class 2 master for the traditional engineering station.

Alcalá, Powell 3 In a typical setup, one Class 1 master would talk to all the slaves on the network using master-slave communications while a Class 2 master would be an Engineering station.

3 Alcalá, Powell 3 In a typical setup, one Class 1 master would talk to all the slaves on the network using master-slave communications while a Class 2 master would be an Engineering station. The two masters would use the token ring algorithm to co-ordinate who could talk on the bus and when. There are watchdog timers built-in to PROFIBUS to ensure that all the communications occur within the required period of time. Figure 2 Master Types, Cyclic vs Acyclic data One of the strengths of PROFIBUS is that its data packet allows for a fair bit of control data, and 244 bytes of data. There are two types of communications channels:. 1. Cyclic This has to occur every bus cyclic, and must be completed within that cycle. This is how the input and output data is communicated. 2. Acyclic This does not have to occur every bus cycle, and can take several cycles to complete. It is used for device parametrization and extended diagnostics. All the cyclic data is defined in a file called a General System Data (GSD) and is used by a Class 1 master to setup the slave on the network. All the acyclic data is defined in a file called an Electronic Device Description (EDD) and is used by a Class 2 master to parametrize and troubleshoot the slave. To join PROFIBUS DP to PA, a device called a PROFIBUS DP/PA Link module is used. It creates a subnetwork under the main DP network and is essentially a PROFIBUS DP slave on one side and a PROFIBUS

4 Alcalá, Powell 4 PA master on the other. PROFIBUS PA runs on MBP physical layer that runs at kbps compared to PROFIBUS DP which typically runs at 1.5 meg. For process control, it is important to know how fast your data is being updated. The PROFIBUS PA update rate depends on how the number of devices, and how much data each of these devices has. Basically the first five bytes of data from each slave take 15.5ms. Each additional five bytes of data from that slave only increases the scan by two ms. In addition, there is 2 ms of overhead. For example if you had five instruments each with five bytes of data, then your PA scan would be 5x = = 9.5ms. Figure 3 PROFIBUS Cycle Time THEORETICAL VS REAL CAPACITY OF THE NETWORK PROFIBUS is a work horse that can theoretically and practically be loaded down to an extreme level. However, just because you can, doesn t mean you should. Keeping loading at a conservative level can be wise in the long run, as this leaves room for expansion to be able to meet both control and user requirements. In this design, the two limits looked at were in connection to the use of the Y link and the DP/PA link. The Y-Link is used for connecting non-redundant PROFIBUS nodes to the redundant cable. It is essentially a DP slave on one side with a built in switch for the redundancy and a DP master on the other. Normally with PROFIBUS DP, the data size is not a big concern. Having each slave have 244 bytes of input data and 244 bytes of output data is not a problem; PROFIBUS can handle it. However, when using the Y-Link, the size of the data downstream of it, is important because you only have a total of 244 bytes of input and another 244 bytes of output. Again, the protocol can handle using all the data, but when designing a new network, is it wise to use up all the space at the beginning? What if new devices need to be added after

5 Alcalá, Powell 5 the Y-Link? What if more data than you thought is needed? For these reasons, it is best to limit the data usage to under 2 bytes each way. In most cases we tried to keep it between 1 and 15 bytes. The DP/PA Link module is used for connecting field instruments like pressure sensors to PROFIBUS. It creates a sub-network for the instruments and like the Y-Link, is a PROFIBUS DP slave on one side and a Master on the other. Also, like the Y-Link, the DP/DA Link module has the same limit of 244 bytes of inputs and 244 bytes of outputs and can handle being fully loaded. However, in this case, there are two design requirements that need to be considered. The first consideration is the speed of response of the instruments for the control requirements. Basically, for some loops, you need the data updated within a certain time period so, if you completely load the network, this will be a problem. Second, your configuration and troubleshooting software such as SIMATIC PDM, uses acyclic communications to talk to the instruments and acyclic communications can only occur after all the cyclic communications have occurred. This means that if you really load up the PA bus, then this software will be really slow and could, for all practical purposes, be un-usable. For these reasons, this design tried to keep the number of instruments per Link module down to between 15 to 18 even though it can handle up to 64. LARGE SCALE NETWORK VS CHALLENGE FOR ALL SMART NETWORKS There are different ways of defining a small, medium, and large network. The simple definition is based on the number of devices: Small: 5 < Devices <1 Medium: 1 < Devices < 35 Large: 35 > Devices An alternate definition is that a large network is a network where the boundaries of the network can be exceeded. In the case study example, the number of devices does exceed 35 devices. Also, as the design progressed, it certainly became apparent that if we were not careful, the boundaries of the requirements could be exceeded. In general, it came nowhere near exceeding the boundaries of the network. PROFIBUS is capable of handling a lot more than this example. However, since this application had smart transmitters and control was involved, hard limits were there and, without thinking about the design, it would have been easy to have exceeded these limits.

6 Alcalá, Powell 6 APPROACH TO DESIGN ITERATION In order to establish a route map for the design of PROFIBUS large networks, it is recommendable to divide the design phases in three iterations. These iterations can be scheduled to execute in a typical project delivery process of 3%, 6%, and 9%. 1. Iteration 1 The focus for this phase is the definition of major components based on the system architecture. For this, the order of ideas should be as follows: a. Outline the relationship of the PROFIBUS networks with the system architecture. Define the network topologies, and have a clear understanding of the devices and instruments that will be part of the PROFIBUS DP and PA network. b. Define the mayor hardware components that will be the link between the smart devices and the PLCs/RIOs. c. Outline the PLC and RIOs locations and its relationship with the process areas to be controlled. d. Develop a design criteria for network segmentation based on the following variables: i. Device s physical location ii. iii. Device s count Byte Counts e. Procure the evaluation of a third part consultant and experienced vendor to provide input to the system architecture, mayor components, and design criteria defined during this phase. 2. Iteration 2 The objective is to develop network details layout. Based on the design criteria established in the previous iteration the network can be segmented including the evaluation of the following variables that may affect the communication performance: a. Networks length b. Current limitation c. Cycle times 3. Iteration 3 The goal is the evaluation of the network flexibility for changes during the system life cycle, which may include changes during construction. Establish a criteria of installed spare capacity for new instrumentation on existing segment or new segments is relevant as part of the O&M documentation to be delivered to the client.

7 Alcalá, Powell 7 EVALUATION OF NETWORK SPECIFICATION VS. THE PROJECTS REQUIREMENTS The network design must satisfy both the project requirements and the network specification. The first elements to consider are the Project Scale and the performance requirements. At the same time, the study of PROFIBUS specifications will establish the foundation for the design criteria. Project Scale and Performance Requirement The case study used for this research was the Vitag Biosolids Fertilizer project in Zellwood Florida, a new fertilizing plant with the following scales and performance requirements: 1. A total of 198 PROFIBUS DP DEVICES organized according to the following structure: a. MCCs: 3 i. VFDs: 62 ii. Starters: Metering pumps: 1 3. Valve actuators: 19 iii. Power metering devices: 4 4. PROFIBUS DP Instruments: RIOs: 1 a. Hardwired AI/AO: 17 b. Hardwired DI/DO : 39 c. PROFIBUS PA INSTRUMENTS: Required Up time: % 7. Continuous control loops: Cyclical communication of more than 15 I/O points 9. Acyclic communication of more than 1 tags to be monitored and configures by Asset Management Software 1. Cycle time < 5 ms for ON/OFF control, < 3 ms for PID control 11. System Architecture and Wiring topology

8 Alcalá, Powell 8 a. Redundant PROFIBUS DP and PLC Architecture b. Area to be networked: 16 thousand square feet of process c. More than 15 feet of networked to be wired d. Process Areas: 5 e. Classified areas: Change management and Expansion a. Required the addition of new devices with minimal to no operation interruption b. 2% of actives network spare required The parameters of the network specification that were taking in account to define the design criteria are the following: PROFIBUS DP 1. Max In/Out Bytes per station: 244 bytes 2. Number of Devices per segment: Number of segment : 1 4. Number of stations addresses: Segment Length vs Baud Rate (Apply only on Copper) 6. Maximum Amount of repeaters: 9 PROFIBUS PA 1. Fixed baud rate: kbps 2. Maximum number of Slave: Number of segment: 1 4. Number of stations addresses: Power Limitation 6. Segment Cable length depends on coupler used

9 Alcalá, Powell 9 DESIN CRITERIA PROFIBUS DP 1. Maximum data after the Y-Link: Less than 2 bytes 2. Maximum Number of PROFIBUS Device in one segment: % of active network space available for expansion PROFIBUS PA 1. Maximum data after the IM: Less than 2 bytes 2. Maximum Number of PROFIBUS Instruments in one segment: Maximum current capacity to be used based on type of coupler: 6% 4. 2% of active network space available for expansion. SYSTEM ARCHITECURE Based on the project requirement, a high availability system was defined as the main control component of the system. To satisfy the high availability system, a redundant PROFIBUS DP network was installed; running over Fiber optic across three main segments for the RIOs and one redundant copper segment for the MCCs that will be located in the electrical room where the Main PLC reside. See the figure below. Figure 4 Case Study System Architecture There was also a requirement to connect a number of non-redundant PROFIBUS DP segments onto the redundant network. This was done with two Y-link modules. For PROFIBUS PA segments, two Interface Modules IM with the capacity to hold up to five DP/PA couplers was used for a similar purpose.

The System architecture requires a redundant PROBIBUS DP backbone where each Remote I/O uses a Y-Link module that

10 Alcalá, Powell 1 Figure 5 Redundant System to not redundant Segments The figure below shows how the PROFIBUS DP Network segmentation was defined to collect data from two MCCs. The System architecture requires a redundant PROBIBUS DP backbone where each Remote I/O uses a Y-Link module that allow the interconnection of none redundant devices into a Redundant PLC architecture. Devices 13 Devices 17 Devices 17 Devices 19 Devices 24 Figure 6 Case Study MCCs Segmentation

11 Alcalá, Powell 11 Following the design criteria in the first iteration of the project, the MCCs were organized in five segments. A further evaluation of the MCCs shop drawing requires a revision of the network segmentation to match the spare capacity of the MCCs. The following pictures show the impact of the vendor MCCs network. Based on the evaluation on the impact to the design criteria a total of seven segments were include for this application. PROFIBUS DP Max In/Out Bytes per station: Less than 2 bytes Number of Devices per segment: 24 MCC1 Total Devices VFD MCC1 Y Link 1 Bytes 38 FVNR RVSS PM IM Bytes Figure 7 Case Study MCCs Segmentation Although the order of the MCC cabinets that contain VDFs, Reverse starters, and Singles starter is basically defined by the vendor, it is useful to understand the best combination of VFD and starter that maximizes the use of the segments, and meet the design criteria. Having in account that a MCC power monitor requires at least 4 bytes, and that the average byte count for a VFD is 12 Bytes and 4 bytes for a singles starters, the subsequent mathematical model was built. X number of starters with an average of 4 byte used Y Number of VFD with an average of 12 Byte used 4 bytes used by Power monitor 4X + 12Y + 4 < 2 bytes X+ Y < 23

12 Alcalá, Powell 12 The combination of VFD and Starter that meet the design criteria are 8 and 12. Based on the network evaluation and the design criteria, the segmentation for the MCC 1 was changed as is shown in the below figure. The MCC1 in total have capacity for 45 PROFIBUS DP connections from which there will be 38 active devices. This leaves 15% spare capacity which will be enough for future expansion within the capacity of the MCC. MCC1 MCC1 Y Link 1 From 1F-C5 To 5-K2 Total Devices VFD FVNR RVSS PM Bytes MCC1 Y Link 2 6-H21 8F-A3 17 Bytes MCC1 Y Link 3 9F-A31 Bytes 12F-E43 13 IM Bytes Figure 8 Revise MCC 1 Segmentation

13 Alcalá, Powell 13 RIO/PLC LOCATION AND SEGMENT CALCULATION The first step to design the network segments layout is to correlate the RIO/PLC location versus the instrument that will be mounted in the process area (See figure below). The second step is to define the segments routing through the different group of instruments (See figure below). A simple approach during the design phase for PROFIBUS PA is grouping the instruments around one junction box. The specification may be written to require the network installer to submit the plan of the network routing for approval. This information may be use to verify the voltage drop and current consumption for each segment. Preliminary calculation of voltage drop and current consumption are important, allowing reserve for changes related to the finals installation. For PROFIBUS DP, an analysis of the segment length to estimate the baud rate is recommended. Figure 9 Case Study RIO Location With the location and datasheet for each instrument of the segment, calculation can be executed using different tools available in order to verify critical electrical parameter for correct operation. See figure 1.

Alcalá, Powell 14 FDC 157 L Input Check R Value Result 43 m Max.

(except for AFDiS) Segment Calculation Value Result Required Current (L-->R) 146.

6 V Minimal AFD Input Voltage (R-->L) Startup Surge Current (L-->R) 146.

Distance 5. m Capacitor Charge Current (L-->R) 146.

14 Alcalá, Powell 14 FDC 157 L Input Check R Value Result 43 m Max. Spur Lengths Spur Current Limits 3 V Segment Length* 1 ma Device Count 4 Spur Count (>1m) 4 * Trunk + Spurs (except for AFDiS) Segment Calculation Value Result Required Current (L-->R) 146. ma Required Current (R-->L) -- Minimal AFD Input Voltage (L-->R) 29.6 V Minimal AFD Input Voltage (R-->L) Startup Surge Current (L-->R) 146. ma Startup Surge Current (R-->L) Startup Min. Voltage (L-->R) Ring. m AFD Type AFD4AFD4 5. m Rel. Distance 5. m Capacitor Charge Current (L-->R) 146. ma Capacitor Charge Current (R-->L) Capacitor Charge Min. Voltage (L-->R) 29.6 V Capacitor Charge Min. Voltage (R-->L) Coupler Switch Current Loop Stability Check Reset V Startup Min. Voltage (R-->L) Calculate -- AFD4AFD m AFD4AFD4 m 2. AFD4AFD4 m. AFD4AFD4 m. 2. m Figure 1 Case Study Bus calculation

15 Alcalá, Powell 15 SEGMENTATION TABLE Once the list of the RIO/PLC vs Instruments locations have been outlined in a spreadsheet, a summary of the segment listed to compare the variables defined as critical in the design criteria. A revision of the bytes frame to be read and written and current consumption for the PROFIBUS PA instruments are typically found on the Instrument documentation. The Cycle time will also be calculated in order to formulate a general overview of the network performance. An example summary table is shown below. RIO Segment Devices Bytes Current Average Scan Average Byte Consumption Time Second Weighted Scan ma Time Second MAIN PLC MAIN PLC MAIN PLC MAIN PLC MAIN PLC GPCP RIO1 GPCP RIO2 GPCP RIO2 GPCP RIO2 GPCP RIO3 GPCP RIO3 GPCP RIO3 GPCP RIO3 GPCP RIO3 GPCP RIO4 GPCP RIO4 GPCP RIO4 GPCP RIO5 GPCP RIO5 OCCP_RIO1 OCCP_RIO2 OCCP_RIO2 Eroom DP/Y-Link 1 Eroom DP/Y-Link 2 Eroom DP/Y-Link 3 Eroom DP/Y-Link 4 Eroom DP/Y-Link 5 DP/PA Coupler 3 1/3 2 DP/PA Coupler 3 1/3 2 DP/PA Coupler 3 1 DP/PA Coupler 4 1 DP/PA Coupler Ex(i) 3 1 DP/PA Coupler Ex(i) 3 2 DP/PA Coupler Ex(i) 3 3 DP/PA Coupler Ex(i) 3 4 DP/PA Coupler 5 1 DP/PA Coupler 6 1 DP/PA Coupler 7 1 DP/PA Coupler Ex(i) 3 1/3 5 DP/PA Coupler Ex(i) 4 1/4 3 DP/PA Coupler Ex(i) 3-1/3-3 DP/PA Coupler Ex(i) 3-1/3-5 DP/PA Coupler Ex(i) 4-1/ Figure 11 Case Study Segmentation Table

Alcalá, Powell 16 BEST PRACTICES 1. Follow the design rules. There are not a lot of them for PROFIBUS, but they are important. 2. Be conservative in your design and leave room for surprises.

16 Alcalá, Powell 16 BEST PRACTICES 1. Follow the design rules. There are not a lot of them for PROFIBUS, but they are important. 2. Be conservative in your design and leave room for surprises. PROFIBUS networks can be stretched to the limits, but this is not a good practice for the future. Networks tend to grow after the initial design and commissioning. If you push the design too much at the beginning you may run into problems later when you need to expand. 3. Accept that design is an iterative process. Do not try to get it perfect on the first try, and know that things will change. 4. Use a second expert to help out. Many companies have a policy that for major business decisions, you must have a second person signing a second pair of eyes to make sure that everything is correct. This idea was used here with the design, as CDM Smith used Siemens to review the proposal. During this process, mistakes were caught and new discussions started and the design was significantly improved. CONCLUSION Designing any network has it challenges. The design of this one was fairly straight forward. The knowledge to design and implement even a large scale PROFIBUS network is out there and well documented. Making use of that knowledge by using all the resources at your service is key to having a successful project of any size. When dealing with a large project, a detailed evaluation of the network specification verses the projects requirements is imperative. ABOUT THE AUTHORS Francisco Alcalá is a member of ISA and an Automation Specialist for CDM Smith. He has a BSEE from Universidad de Oriente Venezuela and an Operation Management Specialist from IESA Venezuela and Florida EIT. Francisco has 22 years of experience in Instrumentation and Control design, integration, and maintenance in the water/wastewater, petrochemical, and beverage industries. Contact: alcalaf@cdmsmith.com James Powell, P.Eng. works for Siemens Canada (formerly Milltronics) as a Senior Product Specialist in Industrial Communications. He has 28 years of experience in automation specializing in various communications systems including PROFIBUS, PROFINET, Foundation Fieldbus, Modbus, Modbus TCP/IP and HART. In 29 he coauthored a book on PROFIBUS, called Catching the Process Fieldbus, An Introduction to PROFIBUS for Process Automation. Contact: james.powell@siemens.com.

Basics of setting up. PROFIBUS-PA networks. Martin Ruck Siemens. Jason Nicholl Phoenix. Kris Hardaker United Utilities

Basics of setting up. PROFIBUS-PA networks. Martin Ruck Siemens. Jason Nicholl Phoenix. Kris Hardaker United Utilities W05 Basics of setting up PROFIBUS-PA networks Martin Ruck Siemens Jason Nicholl Phoenix Kris Hardaker United Utilities Basics of setting up a Profibus PA Network Agenda 1. What is Profibus PA 2. Benefits