Download presentation
Presentation is loading. Please wait.
Published byLillian McDaniel Modified over 10 years ago
1
Università di Genova Facoltà di Ingegneria Genova 28 settembre 2007 A benchmark for SoftPLC Technology in Windows Environment Valerio Malerba Università degli Studi Genova Facoltà di Ingegneria Chiar.mo Prof. Ing. Paolo Pinceti Dott. Ing. Micaela Caserza Magro
2
Università di Genova Facoltà di Ingegneria Whats a benchmark Genova 28 settembre 2007 A benchmark is a sequence of operations allowing to obtain a performance assessment of a system. A program which evaluates the capability of a computer to perform additions by counting the number of operations executed in a while.
3
Università di Genova Facoltà di Ingegneria Benchmark: requirements Genova 28 settembre 2007 A good benchmark satisfies the following requirements: Linearity Reliability Repeatability Easiness Independence
4
Università di Genova Facoltà di Ingegneria A matter of measure Genova 28 settembre 2007 A benchmark performs a measure. For a measure the following parameters are important: Accuracy Precision Resolution
5
Università di Genova Facoltà di Ingegneria PLC (1) Genova 28 settembre 2007 PLC (Programmable Logic Controller): is an industrial computer that runs periodically three phases aiming to control a manufacturing process. Read input: the PLC reads the incoming variables from the system, such as levels, pressures and temperatures. Elaboration: the PLC elaborates, following a program, the input and produces a new series of values or actions. Update output: the PLC updates its output variables.
6
Università di Genova Facoltà di Ingegneria PLC (2) Genova 28 settembre 2007 The PLC is a machine substantially… HARD REAL TIME
7
Università di Genova Facoltà di Ingegneria SoftPLC Genova 28 settembre 2007 A SoftPLC is instead a PLC completely made via the use of a common personal computer. Despite of a traditional PLC, it offers some advantages: Higher intergrability. Possibility to write more complex programs. Lower cost.
8
Università di Genova Facoltà di Ingegneria PLC and SoftPLC: languages Genova 28 settembre 2007 PLC programs follow a standard, IEC 61131-3, which defines 5 languages: Instruction List Function Block Structured Text Ladder Sequential Flow Chart
9
Università di Genova Facoltà di Ingegneria PLC and SoftPLC: difference Genova 28 settembre 2007 The behaviour of a SoftPLC deeply depends by the operating system on which it runs: if the operating system is not a real-time one (e.g. Windows XP), the SoftPLC isnt either… HARD REAL TIME
10
Università di Genova Facoltà di Ingegneria Purpose of the benchmark Genova 28 settembre 2007 The goal of the benchmark is hence to assess the SoftPLC perfomances under the point of view of regularity of working. A second important purpose is to provide a tool to be used during the projecting phase of an industrial plant.
11
Università di Genova Facoltà di Ingegneria Used applications Genova 28 settembre 2007 These a re the main applications used during the realization and use of the benchmark; Infoteam OpenPCS 2006Infoteam SmartPLC/OPC
12
Università di Genova Facoltà di Ingegneria Used applications Genova 28 settembre 2007 ClientOPC These a re the main applications used during the realization and use of the benchmark;
13
Università di Genova Facoltà di Ingegneria OPC Genova 28 settembre 2007 OPC (OLE for Process Control): is a series of standards based on Microsoft DCOM regarding the communication among devices in an industrial network (sensors, PLCs, etc.). OPC
14
Università di Genova Facoltà di Ingegneria The network Genova 28 settembre 2007 Complessively, the network which is affected by the benchmark is the following:
15
Università di Genova Facoltà di Ingegneria The benchmark (1) Genova 28 settembre 2007 The main component of the benchmark is a SoftPLC program that measures its own times, allowing to get information about the overall performances. The language chosen for this duty is Instruction List. The reason is due to its low level programming and therefore its speed and lightness.
16
Università di Genova Facoltà di Ingegneria The benchmark (2) Genova 28 settembre 2007 The program which gets the measure can be represented with a finite state machine:
17
Università di Genova Facoltà di Ingegneria Clock resolution Genova 28 settembre 2007 The GETTIME function implemented by OpenPCS to detect the times strictly depends by the so-called Windows clock interval.
18
Università di Genova Facoltà di Ingegneria The time packet Genova 28 settembre 2007 The time value is recorded inside a 32 bit variable with a resolution of 1ms.This supplies an overmuch range, which can be optimized by transforming the 32 bit variable into a true packet with three decimal fields. Number of cycles: its the number of the current cycle. Empty field: its a value that is kept at zero in order to improve the readability. Time value: is the time elapsed since the program was launched.
19
Università di Genova Facoltà di Ingegneria Experiments Genova 28 settembre 2007 During this thesis numberous tests have been conducted, under different conditions. Considered interesting for study are the cases with different processor occupation levels. Furthermore, being each element in the network a Windows process, various combinations of priority have been considered. In order to generate the occupation levels of the processor, a small software has been developed, called CPU Waster.
20
Università di Genova Facoltà di Ingegneria Metrics Genova 28 settembre 2007 These are the used metrics to evaluate the performances of the network: Mean period: is the average duration of a SoftPLC cycle. Minimum and maximum: minimum and maximum duration of a PLC cycle; in particular, the maximum provides the worst case. Variance: is the variance of a SoftPLC cycle. Lost cycles: is the percentage of cycles which arent arrived to the OPC client. SoftPLC Server OPC
21
Università di Genova Facoltà di Ingegneria Results (1) Several preliminar tests have been made to measure the CPU occupation of each process involved in the network. The most interesting case is the OPC client: Server OPC with low priority Server OPC with high priority
22
Università di Genova Facoltà di Ingegneria Results (2) The percentage of cycles lost by the OPC server increases with the the load but only when its priority is lower or equal to the other processes ones: Server OPC with high priority Server OPC with low priority
23
Università di Genova Facoltà di Ingegneria Results (3) The mean duration of a PLC cycle is almost constant with any load but changes a little depending by the combination of priorities: SoftPLC con high priority
24
Università di Genova Facoltà di Ingegneria Results – Timer mode
25
Università di Genova Facoltà di Ingegneria Results – Other macchines Worthly of mentions are some properties due to different clock intervals on other machines:
26
Università di Genova Facoltà di Ingegneria Conclusions and future developments (1) The tests made show that, under proper conditions, the SoftPLC can satisfy the requests of a control system. Multi-core architecture Higher number of OPC clients Altered update rate for the OPC server Routine that weighs down the SoftPLC code Adaptation of the benchmark to other platforms, different from OpenPCS
27
Università di Genova Facoltà di Ingegneria Conclusions and future developments (2) The benchmark can be extended to evaluate the performances of other elements in the network:
28
Università di Genova Facoltà di Ingegneria THANK YOU!
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.