Sie sind auf Seite 1von 4

_______________________________________________________________________________

8th IWA Symposium on Systems Analysis and Integrated Assessment

Virtual commissioning of process automation software in


drinking water treatment plants

G.I.M. Worm*/**, A.W.C. van der Helm***/**, P. Nijdam***, K.M. van Schagen****/*****, L.C. Rietveld**
* PWN Water Supply Company North-Holland, PO Box 2113, 1990 AC Velserbroek, the Netherlands
(E-mail: ignaz.worm@pwn.nl)
** Delft University of Technology, Faculty of Civil Engineering and Geosciences, Department of Water
Management, PO Box 5048, 2600 GA, Delft, the Netherlands
(E-mail: l.c.rietveld@tudelft.nl)
*** Waternet Amsterdam Water Supply, PO Box 94370, 1090 GJ, Amsterdam, the Netherlands
(E-mail: alex.van.der.helm@waternet.nl; pieter.nijdam@waternet.nl)
**** DHV Water, PO Box 1132, 3800 BC, Amersfoort, the Netherlands
(E-mail: kim.vanschagen@dhv.com)
***** Delft Centre for Systems and Control, Delft University of Technology, Mekelweg 2, 2628 CD Delft, the
Netherlands

Abstract
In industry, the traditional factory acceptance test (FAT) of process automation (PA) software is being replaced by
virtual commissioning (VC). In this paper VC is introduced in drinking water treatment. A quantitative measure has
been defined to compare the shortenings of time-to-market of projects tested with VC and FAT, while taking into
account the size and complexity of the software. The method has been applied to five projects with VC and five
projects with FAT. Preliminary results show VC revealed errors which had been undetected by a previous FAT and
down-time during SAT as planned.

Keywords
Virtual commissioning; drinking water treatment; factory acceptance test; process automation emulation; process
model

INTRODUCTION
As in other industries the level of automation has increased up to the level of remote multi-task supervisory
control (Sheridan, 2002). New process automation (PA) software or software updates are tested extensively to
prevent dangerous situations during or after implementation and to limit process disturbances and down-time.

Traditional PA software testing consists of a factory acceptance test (FAT) followed by a site acceptance test
(SAT) in the plant. To start the FAT, the new software is uploaded to a physical test-programmable-logic-
controller (PLC) in an offline environment. The input/output (I/O) signals are simulated with physical switches,
or with a tailor made tool. The possibilities to connect test-PLCs mutually are limited. A single man-machine
interface (MMI) client is available for navigation. During the SAT the plant is out of operation or operated
manually. The FAT aims to minimize the SAT time, by limiting the risks of unexpected or undesired situations.
A new development in the process automation software engineering is VC in stead of the traditional FAT. VC is
the testing of software in a near-reality situation, using multiple virtual PLCs, multiple MMI-clients possibly
covering different hierarchical automation levels, emulated PA software and dynamic virtual I/O. A possible (but
not necessary) option to give I/O dynamic behaviour is by connecting process models to relevant I/O (Worm,
2010). Figure 1 shows the parallels between the PA-system which is connected to the sensors and actuators in
the field, and the virtual commissioning system existing of PA-system emulation and process simulation. The
differences between VC and the traditional FAT are summarized in Table 1.

_______________________________________________________________________________
830 Watermatex 2011: Conference proceedings
_______________________________________________________________________________
8th IWA Symposium on Systems Analysis and Integrated Assessment

Figure 1. VC is a near realistic representation of the PA-system1.

VC FAT
PLC Multiple virtual Single physical
Test scope A drinking-water-treatment-plant A single PLC
Testing of communication Yes No
between PLCs
Testing of functions with Yes No
multi-PLC interactions
MMI clients Multiple, possibly over different Single
hierarchical levels
Virtual I/O Standardized, dynamic Tailor made, static
Modification of software Negligible Significant
before or after testing

Table 1. Differences between virtual commissioning (VC) and the traditional Factory acceptance test (FAT)

1
Visualization by DHV, Waternet and PWN

_______________________________________________________________________________
Watermatex 2011: Conference proceedings 831
_______________________________________________________________________________
8th IWA Symposium on Systems Analysis and Integrated Assessment

The current selection of FAT or VC is mainly determined by the possibilities and restrictions of existing PA
engineering systems and not by final effect on the drinking water production and distribution. The choice should
be made by weighing cost and benefits, see Figure 2. In practice, the costs of VC are easier to determine than the
benefits.

Figure 2. Expected added value of virtual commissioning (Reinhart, 2007)

METHODS AND MATERIALS

Measuring shortening of time-to-market of VC


To measure the shortening of time-to-market, the dimensionless parameter ‘Test-effort’ has been defined. The
calculation of the software-size [hours], including its complexity is (1):

Expected costs
Software size = (1)
Average rate of the software engineers

The efforts [hours] of FAT and VC were expressed by the sum of i) the number of hours spent to FAT/VC by
software engineers and end-users, ii) the number of hours spent to SAT by software engineers and end-users and
iii) the number of hours spent to fix errors within one year after the SAT by software engineers. Data were
derived from the companies’ project administration, SAT-logbook and urgent-alarm registration. To compare
FAT and VC the sums of efforts divided by the software size, were determined, yielding a dimensionless value
for the effort of testing (2). Additionally, to get a qualitative impression, end-users have filled out questionnaires
to evaluate each SAT session.

Duration of FAT / VC + Duration of SAT + Time to fix errors


Test effort = (2)
Software size

The comparison of FAT and VC


Using the measure of Test-effort, five PA-software development projects tested with VC were analysed and
compared with five FAT-tested projects. The costs for the licenses and set-up of the VC test-platform were
assigned to the projects pro rata of the duration of their VC or FAT.

Synergies, advantages of VCp


To realise VCp, two versions of a hydraulic model, one in Flownet, one in EPAnet (Worm, 2009), were
connected to the relevant virtual I/O for the commissioning of the new control of a pellet softening plant.

RESULTS AND DISCUSSION


The method is applied in the first quarter of 2011, the results will be published in the full paper. The first three
projects tested with VC within PWN yielded SATs without unexpected errors or delays, which gave reason to
double the number of licenses of the VC-platform. The set up of ad-hoc black box models for VCp requires

_______________________________________________________________________________
832 Watermatex 2011: Conference proceedings
_______________________________________________________________________________
8th IWA Symposium on Systems Analysis and Integrated Assessment

efforts while being inferior compared to models which are calibrated and validated for support in plant operation
and plant optimisation. Two synergies of VCp are expected when applied in drinking water treatment. By
connecting the existing stand-alone simulation environment to the PA-emulation, the gate will be opened to
high-fidelity training and assessment of operation supervisors. A second synergy is the use of the system for
process optimization by process-technologists and control engineers.

CONCLUSIONS
In this paper VC is introduced in drinking water treatment. A measure was defined to compare the shortenings of
time-to-market of projects tested with VC and FAT, while taking into account the size and complexity of a
specific software project. By doing so, a contribution was made in answering the Efficiency Measuring Problem
of Virtual Commissioning. A first review of the results showed VC revealed errors which were not detected
during a previous FAT. SATs after VC contained no unexpected errors or delays, so down-time was conform
planning. The efforts of testing were less with VC compared to the traditional FAT.

REFERENCES
Reinhart, G. and Wünsch, G. (2007). Economic application of virtual commissioning to mechatronic production
systems. Prod. Eng. Res. Devel., 1(4), 371-379.
Worm, G. I. M., Mesman, G. A. M., van Schagen, K. M., Borger, K. J. and Rietveld, L. C. (2009). Hydraulic
modelling of drinking water treatment plant operations. Drink. Water Eng. Sci., 2(1), 15-20.
Worm, G. I. M., van der Helm, A. W. C., Lapikas, T., van Schagen, K. M. and Rietveld, L. C. (2010).
Integration of models, data management, interfaces and training support in a drinking water treatment plant
simulator. Environ. Model. Softw., 25(5), 677-683.

_______________________________________________________________________________
Watermatex 2011: Conference proceedings 833

Das könnte Ihnen auch gefallen