newly installed GE MBC software


If I understand the TLA (Three-Letter Acronym) in your post someone at your site has volunteered to be a test site by having the new Model-Based Control software running a turbine at your site. There is PRECIOUS <b>LITTLE</b> documentation on this for several reasons. First it's new. Second, it's a work in progress. Third, GE wants to keep the MBC a proprietary secret.

You really should be working with GE or the packager of the unit to understand and resolve the problem.
What alarms relative to the input signal processing are being displayed?

As far as the noise question, the fidelity of the model used in the Model Based Control depends, in part, on the accuracy of the input signals used by the model. The revised Input Signal Processing uses new fault detection algorithms to maintain the accuracy of the model. The amount of signal noise is one of numerous characteristics used in the input signal fault detection and accommodation. The baseline noise levels for each sensor are determined during initial MBC commissioning, and the constants which define the baseline are stored in the controller's non-volatile memory. This baseline determination is the sensor noise training you refer to.

You seem to be in the "Inner Circle" of knowledge on GE MBC. This is a good thing to know.

How is the average technician going to be able to troubleshoot problems with units controlled with MBC? Is there going to be extensive 'Help' screens to aid in understanding problems such as this one?

Are the "alarms" going to have 'Help' screens available to assist with troubleshooting and resolution?

What happens to turbine operation when sensors like this one (supplying information to the (Natural) Gas Fuel Flow calculation) fail or are outside of limits?

Is it intended that some Remote Monitoring and Diagnostics Center will "connect" to the Speedtronic turbine control and diagnose the problem and provide assistance to the on-site technicians?

Thanks very much for your insight and assistance with our understanding of this new development.

When I checked this thread and re-read your post, I realized you used the word "tuning". I should point out that there is a sensor tuning procedure for a few sensors which have a model estimated output used as a backup to the actual sensors. This is a different process than sensor "training", which sets the noise baseline. (The sensor "tuning" process is also performed at MBC commissioning, and adjusts the sensor model estimates based on actual operating parameters.)

I'm not in any particular circle, but have seen a couple of these MBC modifications. MBC is becoming more commonplace, since it's the basis for some additional "features" such as auto DLN tuning and ETS ("enhanced transient stability", i.e., enhanced control system response to grid voltage and frequency disturbances).

In my experience, GEH documents have been made available which explain the changes, and have lists of new alarms and some basic troubleshooting. In some cases, operator training was also done. As you well know, the "pro-activeness" of the organization involved will determine how well information flows and to whom.

One advantage of the model is that it can synthesize some sensors in the event the actual sensors are failed, allowing operation to continue, although sometimes in a degraded (de-rated) manner.

MBC software has dynamic data recorders (DDR) which are specific to the software options installed. The DDRs take relevant pre- and post-fault data when various fault conditions occur. Since the DDR outputs are saved in the same manner as the trip histories, they would be available to M&D personnel via the OSM.