A TEST RIG FOR TESTING HIGH PRESSURE CENTRIFUGAL COMPRESSORS MODEL STAGES AT HIGH REYNOLDS NUMBER P. NAVA, M.PROFETI, M. GIACHI, F.SARRI GE-NUOVO PIGNONE G.P.MANFRIDA UNIVERSITY OF FLORENCE Italia Presented at Venezuelan Gas Processors Association (AVPG) XIV International Gas Convention May 10-12, 2000 Caracas, Venezuela
A TEST RIG FOR TESTING HIGH PRESSURE CENTRIFUGAL COMPRESSORS MODEL STAGES AT HIGH REYNOLDS NUMBER P. Nava, M.Profeti GE-Nuovo Pignone - R&D Dept. M. Giachi, F.Sarri GE-Nuovo Pignone - Centrifugal Compressors Dept G.P.Manfrida University of Florence - Energetics Dept. ABSTRACT In this paper a new facility designed and built to test scaled down models of centrifugal compressors stages is described. The facility has been designed to allow a full reproduction of real conditions. Large impellers can be tested to avoid any scale effect between the model and the real compressor and high pressure tests can be performed to test at high density i.e. high Reynolds number. In addition, all kinds of stages (first, intermediate or last) can be mounted to fully reproduce the full scale conditions INTRODUCTION Performance prediction of centrifugal compressor stages is becoming more and more challenging. The need of strong procedures for stages characterization is one of the most important aspects of advanced modern design. Nowadays, even if very high capacity computers are available for numerical simulation, experiments are still the most reliable way to accurately predict performance. Experiments can be performed on full scale machines or on scaled down models. This second option is widely used [1], [2] during design phase and for research purposes when many options need to be tested to find optimized solutions. To reproduce the real conditions the tests must be carried out following, firstly, the geometrical similarity between the model and the real compressor and following the basic fluid dynamic similitude rules which for a centrifugal machine are expressed in terms of non-dimensional flow coefficient, volumes ratio, Reynolds number and Mach number. If these quantities are reproduced the model is absolutely fully representative of the real machine and the performance measured on the model can be directly extrapolated to the real compressor without any corrections. Two problems arise: sometimes it is difficult to design models which reproduce the real configuration (first or intermediate or last stage) including gaps, seals, wall roughness...second, it is often difficult to test at high Reynolds number as it is typical of some chemical and petroleum applications where narrow stage operating at very high pressure are involved. Reynolds number is the most difficult to be achieved for these applications and many papers have been written in the past to take into account the approximation due to a mismatching of this parameter [3], [4], [5], [6]. Página 2
Recommendations are also given in Standard ASME codes [7] to define the allowable maximum difference between Specified Reynolds number and Test Reynolds number. Anyway, it often happens that specific data are needed for new applications and more detailed information are required even for standard applications to accurately predict compressor performance. Following this scenario a new test rig has been designed and built. The main features of this facility is its flexibility and the possibility of testing all kinds of stages at different test conditions: intermediate as well as first or last stages can be tested with upstream or downstream volute, small size impellers (300 mm diameter) can be tested at high pressure (100 bar) or large impeller (1000 mm diameter) can be tested at low pressure (1-3 bar). In this paper the description of this new test facility is given. GEOMETRICAL SIMILARITY AND SCALE EFFECT Three kinds of stages can be identified for centrifugal compressor applications: first stage with upstream inlet plenum, intermediate stage with upstream and downstream return channels and last stage with downstream discharge volute. Many experiences have shown that the differences in the stationary components may have strong influences in the performance of the whole stage hence it is important to test the stage as it is in the real machine with upstream and downstream proper component. So doing the geometry of the stage is properly reproduced, but it is well known that this geometrical similarity cannot be achieved for all components: seals, gaps, wall roughness... cannot be scaled down proportionally, mainly because of mechanical constraints. For this reason all manufacturers have their own rules to correct the data from small size to large size impellers which take into account the effects of all those details which are different between the machine and the model stage tested in the test stand. REYNOLDS EFFECT The Reynolds number is a ratio between inertial forces and viscous forces. gasdensity refvelocity ref Re y = gasviscosity dim ension This number is defined in a centrifugal compressor stage assuming the gas property at the inlet of the stage, the reference velocity is the impeller peripheral speed and the reference dimension is the exit width of the impeller. Reynolds number is an index of the kind of the flow, which is present inside the stage A low Reynolds number means a predominant effect of gas viscosity and a high Reynolds number means a predominant effect of gas density, gas velocity and impeller size. A 300 mm impeller running at 12000 RPM to compress natural gas at 600 bar may have a Reynolds number of the order of tenths millions because of its high gas density. A large 1400 mm impeller running at 3600 RPM to compress air at ambient conditions may have a Reynolds number less than one million. Página 3
If they are tested at standard conditions using a 400 mm impeller, one clearly understands that in the first case the achievement of the real Reynolds number in the laboratory may be a problem. TEST RIG DESCRIPTION The new facility described in this paper has been designed and built to overcame all the above mentioned items. This facility is shown in figs. 1 and 2: it IA a closed loop aero circuit whose main features are given in Tab.1. A special section of the circuit is dedicated to stage model assembly (fig. 3). Different assemblies are possible to test different stages types: first stage with upstream inlet plenum, intermediate stage with upstream and downstream return channels and last stage with downstream discharge volute. Furthermore, the test rig can be operated in three different configurations to test at different conditions: standard: medium size impellers (400 mm), medium pressure (6 bar max) high pressure: small impellers (300 mm), high pressure (100 bar max) large impeller: large impeller (1000 mm max), low pressure (1 bar) DATA ACQUISITION SYSTEM The Data Acquisition System is fully automatic. Steady pressures are measured up to 512 measuring points (fig. 8) and as well as 80 steady temperatures. Furthermore, 5 channels for unsteady fluctuating pressures are available. Data acquisition is remotely controlled by a Pentium PC, which also controls the test rig set-up. The data acquisition software is presently written in LAB_VIEW, while data reduction is performed by an interactive MATLAB code with visual interface and graphical presentation of the results. These can be thus be made available a short time after the end of the tests. ACCURACY Accuracy of the measurements is guaranteed by a rigorous procedure and a high accuracy instrumentation. Large attention is given to get measurements in stabilized condition. Stage inlet and outlet temperatures are continuously displayed as well as polytropic efficiency and the test conditions are considered stable when a difference less than 0.25% is found between two measurements. CONCLUSIONS Centrifugal compressor stages scale model test has been found a very useful and productive tool for new designs validation and for research and development activity. To achieve reliable information of the real machine performance the test must be performed following accurately the geometrical similarity between the model and the real machine and following the fluid dynamic similarity rules. Corrections are applied to the results when full similitude cannot be achieved. To do this special attentions must be dedicated to the model. Página 4
In this paper a test rig has been described which allow the test of all kinds of stages of different size and at different test conditions to fully match all similarity rules reducing almost completely all uncertainties in extrapolating performance from the model to the real machine. Tests can be run using large size impellers (up to 1000 mm diameter) or high pressure (100 bar). Página 5
REFERENCES 1. Benvenuti E., Aerodynamic Development of Stages for Industrial Centrifugal Compressor. Part 1: Testing Requirements and Equipment - Immediate Experimental Evidence, ASME Paper 78- GT-4 2. Benvenuti E., Aerodynamic Development of Stages for Industrial Centrifugal Compressor. Part 2: Test Data Analysis, Correlation and Use, ASME Paper 78-GT-5 3. Wiesner F.J., A New Appraisal of Reynolds Number Effects on Centrifugal Compressor Performance, ASME Journal of Engineering for Power, Vol.101, July 1979 4. Simon H., Bulskamper A., On the Evaluation of Reynolds Number and Relative Surface Roughness Effects on Centrifugal Compressor Performance Based on Systematic Experimental Investigations, ASME Journal of Engineering for Gas Turbine and Power, Vol.106, April 1984 5. Casey M.V., The Effect of Reynolds Number on the Efficiency of Centrifugal Compressor Stages, ASME Paper 84-GT-247 6. Corradini U., Metodo di Correzione delle Prestazioni degli Stadi Centrifughi Standard per Effetto del Numero di Reynolds, NuovoPignone Int.Rep.CS00054 7. ASME PTC 10, Performance Test Code on Compressors and Exhausters, 1997 Página 6
FIGURES Fig. 1 - Scenografic view of the facility Página 7
Fig. 2 - Facility technical lay-out Electric motor asynchronous, 1500 rpm 2.5 MW Fluid-drive double rotating direction, 50-100% full torque speed range capability Gear box three wheels sets (1500/7500, 1500/15000, 1500/25000) Air/water cooler Pmax =100 bar Tmax=300 C Piping one set for low pressure (up to 30 bar) and one for high pressure (up to 100 bar) Tab. 1 - Facility main characteristics Página 8
(a) (b) Fig. 3 - Model assembly section for low pressure (a) and high pressure (b) tests Página 9