Method for collecting machine vibration data

• US 5,610,339 A
• Filed: 03/06/1996
• Issued: 03/11/1997
• Est. Priority Date: 10/20/1994
• Status: Expired due to Term

• Current Assignee: Ingersoll-Rand Company (Trane Technologies plc)
• Original Assignee: Ingersoll-Rand Company (Trane Technologies plc)
• Inventors: Kirkpatrick, Paul A., Haseley, Robert K.
• Primary Examiner(s): CHAPMAN JR, JOHN E
• Application Number:
  US08/612,196
• Time in Patent Office: 370 Days
• Field of Search: 73/659, 73/660, 73/1 DV, 364/508, 418/2
• US Class Current: 73/660
• CPC Class Codes: F04C 2270/125 Controlled or regulated
• CPC Class Codes: F04C 28/28 Safety arrangements; Monitoring
• CPC Class Codes: G01H 1/003 of rotating machines G01H1/10 takes precedence

Abstract

A method is provided for determining and for predicting the present and future condition of a machine having at least one rotative element. The method includes an initial step of storing in memory a predetermined logic routine and at least one predetermined key frequency of the at least one rotative element of the machine. The machine is operated in a predetermined operational state and, during operation, the mechanical motion of the at least one rotative element is sensed by at least one sensor. The sensed mechanical motion of the at least one rotative element is converted into a corresponding electrical signal, and the predetermined operational state of the machine is correlated with a corresponding electrical signal. The corresponding electrical signal is inputted into corresponding vibration data by utilizing the predetermined logic routine. Then, the corresponding vibration data is compared with the at least one predetermined key frequency to predict the present and future condition of the machine. A control signal is transmitted to the machine, in order to control the machine, in response to the present condition and the predicted future condition of the machine.

Claims

1. A method for predicting and for detecting the condition of a screw compressor having at least one rotative element, the method comprising the following steps:

storing in a memory a predetermined logic routine and at least one predetermined key frequency value of the at least one rotative element of the compressor;

trending, at regular intervals, to determine whether the compressor vibratory condition has changed, said trending comprised of the following steps;

operating the compressor in a predetermined operational state;

sensing the mechanical motion of the at least one rotative element during compressor operation;

converting the mechanical motion of the at least one rotative element into a corresponding electrical signal;

correlating the predetermined operational state of the screw compressor with the corresponding electrical signal;

inputting the corresponding electrical signal to a processor;

processing the corresponding electrical signal into corresponding vibration data of the compressor by utilizing the predetermined logic routine; and

analyzing the corresponding vibration data for trends with the at least one predetermined key frequency value of the at least one rotative element of the compressor and comparing the vibration data with any previously collected vibration data to determine the present condition and predict the future condition of the compressor;

decreasing the length of the regular interval between trending steps if the compressor experiences a negative deviation in the vibratory condition; and

controlling the operation of the compressor in response to the determined present condition and predict the future condition of the compressor by sending a command signal from the processor to the compressor.

2. A method, as claimed in claim 1, and wherein the predetermined logic routine utilizes a fast Fourier transform algorithm to process the corresponding electrical signal into corresponding vibration data.

3. The method for predicting and detecting the condition of a screw compressor having at least one rotative element, as claimed in claim 1 further including the step of calibrating the processor at predetermined time intervals, while the compressor is operating.

4. The method for predicting and detecting the condition of a compressor having at least one rotative element, as claimed in claim 1 further including the step of initializing the processor at predetermined time intervals, while the compressor is operating.

5. The method for predicting and detecting the condition of a compressor having at least one rotative element, as claimed in claim 4, wherein at least one sensor is attached to the compressor, and the processor communicates with a data memory, a program memory and a display interface, and wherein the the method includes at least one of the following steps performed before the trending step:

identifying the compressor to be monitored;

identifying the at least one rotative element to be monitored;

identifying the number of sensors; and

determining the state of the data memory, the program memory, the display interface, and the processor.

6. A method for predicting and detecting changes in the vibratory condition of a screw compressor, which includes at least one rotative element, said at least one rotative element having at least one key frequency, the method for predicting and detecting changes in the vibratory condition of the screw compressor, comprising the following steps:

before permanently installing the screw compressor, collecting first vibration data for the screw compressor at a predetermined screw compressor operating state;

compiling the first vibration data into a corresponding benchmark footprint for the predetermined operating state of the screw compressor and comparing the first vibration data with the key frequency of the at least one rotative element to determine if the screw compressor is in a desired vibratory operating condition;

permanently installing the screw compressor at an operation site;

storing in a memory a predetermined logic routine, which includes a fast Fourier transform algorithm, and at least one of the key frequency value of the at least one rotative element of the screw compressor;

collecting second vibration data for the at least one rotative element at the predetermined compressor operating state;

compiling the second vibration data into a corresponding second benchmark footprint for the predetermined operating state of the compressor and comparing the second vibration data with the stored at least one key frequency value of the at least one rotative element to determine if the compressor is in a desired vibratory condition;

trending, at regular intervals, to determine whether the compressor vibratory condition has changed, said trending comprised of the following steps;

switching the compressor from an actual compressor operating state and operating the screw compressor in the predetermined operating state;

sensing the mechanical motion of the at least one rotative element during compressor operation in the predetermined operating state;

converting the mechanical motion of the at least one rotative element into a corresponding electrical signal;

correlating the predetermined operating state of the compressor with the corresponding electrical signal;

inputting the corresponding electrical signal to a processor;

processing the corresponding electrical signal into corresponding vibration data of the compressor by utilizing the predetermined logic routine;

collecting vibration data of the compressor by storing in a memory the corresponding vibration data; and

comparing the corresponding vibration data with the at least one predetermined key frequency value of the at least one rotative element of the machine and any vibration data previously collected during a previously executed trending step and stored in memory, to highlight trends in the vibratory condition of the machine and thereby determine the present condition and predict the future condition of the compressor;

decreasing the length of the regular interval between trending steps if the compressor experiences a negative deviation in the vibratory condition;

returning the compressor to the actual compressor operating state; and

controlling the operation of the compressor in response to the determined present condition and the predicted future condition

of the compressor by outputting a command signal from the processor to the compressor, while the compressor is operating.

7. The method for predicting and detecting the condition of a compressor having at least one rotative element, as claimed in claim 6, the compressor further including at least one sensor connected to the compressor by at least one sensor connection, the processor including a power supply, the method including the following steps performed before the step of installing the screw compressor, determining the status of the power supply;

determining the excitation voltage of the at least one sensor;

determining the status of the at least one sensor connection and determining the status of the signals.

8. A method for predicting and detecting the condition of a compressor having at least one rotative element, the method comprising the following steps:

storing in a data memory a predetermined logic routine, which includes a fast Fourier transform algorithm;

storing in the data memory, at least one key frequency value of the at least one rotative element for each of the different predetermined operating states of the compressor;

switching the compressor from an actual compressor operating state and operating the compressor in a predetermined operational state;

sensing the mechanical motion of the at least one rotative element and collecting vibration data at predetermined time intervals;

processing the collected vibration data into at least one vibration footprint using the predetermined logic routine for each of the at least one rotative elements;

converting each of the at least one vibration footprints to a corresponding electrical signal and storing each of the corresponding electrical signals within the data memory;

correlating each of the previously stored at least one key frequency values at the operating state, with a corresponding vibration footprint;

comparing each of the at least one vibration footprints with the corresponding at least one stored key frequency value to determine whether each of the at least one vibration footprints is within the acceptable range of the corresponding at least one stored key frequency value of the at least one rotative element and comparing the vibration footprints with the vibration footprints generated during previous time intervals to reveal trends in the vibratory condition of the compressor;

decreasing the length of the regular interval between trending steps if the compressor experiences a negative deviation in the vibratory condition;

returning the compressor to the actual compressor operating state; and

controlling the operation of the compressor in response to the determination whether each of the at least one vibration footprints is within an acceptable range for the compressor by outputting a command signal from the processor to the compressor, while the compressor is operating.

9. The method for predicting and detecting the condition of a compressor having at least one rotative element, as claimed in claim 8 further including the step of calibrating the processor at predetermined time intervals, while the compressor is operating.

10. The method for predicting and detecting the condition of a compressor having at least one rotative element, as claimed in claim 8 further including the step of initializing the processor at predetermined time intervals, while the compressor is operating.