Our solartron 7835 is having erratic behavior while diesel fuel is being delivered. We get stable readings for all other products. however when diesel is being delivered, the density jumps imitating the tail end of the batch, thus signaling the control center to swing the valves. We have tried a few things. I have brought a rep out to look at the meter itself. he gave the all clear and said there was nothing wrong with the meter. Also the pump and motor feeding the have been replaced. Could the barrier be failing etc... I would welcome any ideas or advice.

 

"Erratic behavior" is generic statement which tell us very little about your problem (I.e. short or long duration spikes, multiple spikes, pulses, noise, direction of spikes, consistency (regularity, time dependency), etc). In any case, how about a wild theory that the density meter is reading correctly. If it is possible take samples during the erratic behavior and run them in the lab to see if there is an actual density change. Maybe you are getting light end or non-condensibles in the diesel as they may cause your densitometer to read strangely. Are you blending before transfer (e.g. inconsistent or incomplete blending causing density inconsistencies)?

William (Bill) L. Mostia, Jr. PE
Sr. Consultant
SIS-TECH Solutions, LP
Any information is provided on Caveat Emptor basis.

--------------------------------------------------

 

Do you get water in the diesel? As Bill mentioned, why not draw a sample when the density signal "jumps".

Regards,

Tomy Zacharia

 

Sorry first time posting on a forum. I will try to provide a clearer picture. From the trending history that I can see, it looks like short duration spikes in increased density of around 5-8kg/m3. Also it seems to be happening at the tail end of the batches, however it has happened at the beginning of the batch too. I am trying to find out if there is blending before the batch is sent. There are two other densitometers before the one in question and they read constant when the batch passes by them. The spiking also occurs at lower pressures (550 kpa) and at higher (1600 kpa)I have drawn samples while diesel was being delivered and it's density has been what it is supposed to be. There was no spiking at that time though. It is a remote site so it is difficult to reach in time when the spiking occurs to pull a sample.

 

Are the two densitometers the same technology as the Solartron? Different technologies sometimes respond differently or don't respond the same disturbance so it is still possible that the density is changes.

There are some other possibilities if we assume based on the other two densitometers indicating no spike in density indicate no contaminant present:

1. The signal wires are run next to power wires and something is turning off/on causing an electrical spike in the densitometer signal line. Are there anything electrical turning off/on at the times of the spikes?

2. You have a power quality problem and your getting the spikes from there.

A possible fix if the spikes are short duration is a short delay timer on the shutdown say 1-5 sec which will filter out any short duration spikes (e.g. the signal has to stay longer than 1-5 sec before you will shutdown).

William (Bill) L. Mostia, Jr. PE
Sr. Consultant
SIS-TECH Solutions, LP
Any information is provided on Caveat Emptor basis.
--------------------------------------------------

 

The fact that one densitometer spikes and another doesn't, doesn't even mean the two sensors are responding differently but that they are configured differently and the signals processed differently. Given the complexity of the Solartron software a great many users who didn't budget for commissioning end up doing just enough to get the sensor working well but not necessarily optimally for their application.

For example, the raw measurement could be processed using a simple rolling average mechanism that rejects the oldest value and adds the newest value, then rejects highest and lowest and then provides an average of what remains.

How smooth the signal is depends on how many values remain to be averaged. The more values, the more stable the reading and the less responsive to step changes (there is also a dynamic averaging set up that provides for both stable readings and fast response to step changes but which then takes a while to settle down again after the step change).

Chances are that even if they are all Solartron 7835s they all have different configurations.

But the 7835 software also has some other filters and so too may the other density meters.

For example, there is a time period trap. This is designed to catch transient readings. The spikes, for example.

There are several options here.

In closed loop control the last thing you need is a spike sending your control off kilter as it may respond to the spike and then spend some time coming back on control. So one of the options is for the measured value to be held at the last good value until either the transient has passed (in which case you wont see it but it may appear in the alarm history) or a set time has elapsed after which it defaults to the live value.

But in other applications you need to see the spikes e.g. metering but also alarm so you can increment error count registers.

So many ways the software can be configured which will make two identical sensors on the same duty appear to behave very differently.

So beware of making unfortunate comparisons to other sensors.

The only way to see what is really going on is to data log the raw data. There is an excellent program for this from Solartron called FCCONFIG.

This not only lets you configure all the possible settings but also to make a detailed data log of raw and processed signals. It can even only log data when an alarm is triggered.

Now, the chances of the sensor being faulty and only faulty with diesel are pretty remote.

What you have to consider is that what you are seeing is an artifact of the flow regime which is unique to the diesel flow, assuming that what you meant was that this sensor is used for multiple fluids as in a multi-product pipeline application, and one of the products shows the spikes and the others don't.

Spikes in the data generally mean some extreme transient event.
The options are a step change in density such as when some tramp fluid pocket is flowing or, more likely, an air pocket.
What the display shows may only be that portion of the transient captured in that update or what is left after the filters have processed it.

So I'd have to assume that this is something happening in the flow lines to the fluid. I'd look to see if there are any "events" corresponding to these spikes.

Or I'd look to see if there are any features of the diesel flow system that are different to the other liquid flow systems.

What you may be seeing is a historical event further upstream. In some refinery applications what you see at the sensor is a result of something that happened way back upstream some 8-9hours before.
It might be that diesel has its own pump and the pump needs maintenance - its drawing air in occasionally or there is a faulty valve in the suctions lines somewhere. Whatever. This is the sort of thing you are looking for and to use the sensor more effectively you probably need to start data logging.

 

I am assuming this is the receiving end of a multi-product pipeline.
Clearly, what you are seeing is a feature associated with the diesel and the process conditions associated with diesel. It may be something at your end or something at source.

It may simply be that the configuration settings are optimum for most products but marginal for diesel and that a simple configuration change is justified. It may also be that the settings for this density meter are different to the settings for the others.

For example, as each product flows you are looking for the interface with the next product.

I assume that what you have are some upstream density meters warning of the approach of the interface. These might not be 7835s they may be 7826 or 7828 tuning fork density meters (I have seen them used by them selves in airport fuel depot receiving stations). The tuning forks are the same fundamental principle just less accurate that the 7835s. It is possible they will be 7826 using the same electronics and software as the 7835.

In any event and whichever method for interface detection you use (zones, step change target value etc.), there are some basic settings to look at.

If you have FCCONFIG it makes life easy. Otherwise you have to go step through the menus by hand.

Basically the early software has a time period trap. This is to prevent alarms, analogue outputs, displays etc spiking for normally expected transients. In the early versions about all you could set was the step change limit. In later versions you get to set the duration.

All you really need to do at this point, assuming all Solartron, is to compare the settings one sensor to another.

What you may find is a minor difference that means the settings are marginal for diesel in this sensor and optimal for all other products but optimal for all in other sensors.

The concept behind these settings is to ignore the transients and noise associated with an interface and look for the transition to the new product. The spikes you may be seeing may simply be those that are of longer duration than other density spikes which the filters hide.

Now it may be also that if you are logging data you are logging at a fairly long interval for continuous logging and to see all that is going on you need to set up special short interval logs and to look also at the raw data.

You can use FCCONFIG or any other program. The easy way to do this in FCCONFIG is set up a list of the locations you want and then simply log the list.

There are some settings you can check and compare.

Navigate to
Configure/line density/dens corrections
and check what corrections are selected

Then Nav to:
Configure/line density/line density limits
and check the settings there with attention to noise filter which is step change in time period you have set and trap count, the duration. There to are the hi and low density alarms and a density step limit.

Next navigate to
Configure/line density/fallback
This is the action when the step alarm initiates. It probably ought to be Last Good Value. Meaning that when the step change is detected it hold s the last good value preventing spikes in any displays, alarms, outputs etc. until the trap count period expires.

It may mean that if there is an 8kg/m3 density difference between products and some disturbance at the interface, density alarms alone won’t discriminate between the noise and the actual transition. For the interface to be detected the density must change by the expected amount and the value persist long enough to be accepted as the actual transition and not advanced pockets of the new product.

The value to look at is the trap count. It may be this is more sensitively set for this sensor than the others. It may also mean that you are only seeing some of the noise because the rest is already trapped and not seen.

You might also go to configure/additional features/alarms and look at the physical alarm settings and look for differences.