Will my table design cause future problems

Aaron · Apr 27, 2004

Access/VBA newbie here and I have the following situation:
A research facility mixes elements(LotID), processes them,
evaluates the end product and then uses it in a part which
will also get tested.
The table structures I have so far:
tblComposition (created this way because a Lot# can be
made up of 1-? elements)
LotID
Element
ElementRatio

tblProcessing
LOtID
ProcessEquip
RunDate
%Source

tblEvaluation
LotID
ParticleSize
ECValue
FinalMAss

tblParts
Part#
LotID
Account
Performance

One of the key deliverables from this database is to
provide Evaluation data and Part listings for every
combination of Elements and %Source information. i.e;
LOT#AALL12 had a 10%Source and was made up of Al,Si,Cu at
a ratio of 60,20,20 and was used in PART#1234,PART#5678,etc
If I continue down this path I will need to concatenate
the records in the tblComposition and I don't know how big
a problem that will be. I also don't know what other
problems I am not considering. Please advise. Thanks.

TC · Apr 28, 2004

Aaron, you will get nowhere with this question until you identify what is
the "primary key" of each table, and state exactly what thing or event is
identified by each row in the table.

For example:

tblClient - one row for each client known to the system.
ClientID (PK)
Forname
Surname
etc.

This might help:
http://support.microsoft.com/support/kb/articles/Q100139.ASP

HTH,
TC
(off for the day)

Aaron · Apr 28, 2004

I added some key information, hopefully this along with
other descriptions helps. thanks.

TC · Apr 29, 2004

(snipped)

So:
- each part has one (or several?) tests;
- each part comprises several end products;
- each end product comprises several elements.
Correct?

What is the primary key of that table? What does one row in that table,
represent?

What does one row in that table, represent?

What is the primary key of that table? What does one row in that table,
represent? (I don't understand: "characterization of LotID's after
manufacturing". You need to describe the table in terms of the entitites you
have mentioned so far: parts, tests, elements, end products, etc.)

Aaron, that is the only table that is clearly described so far. Each row in
that table represents a single, unique Part. Each part is identified by a
unique Part#. Part# is the primary key of that table. We need the same level
of understanding of your other tables, before we can comment.

HTH,
TC

Aaron · Apr 29, 2004

-----Original Message-----

So:
- each part has one (or several?) tests? several types of tests
- each part comprises several end products? The combined

elements(LotID) can be used in several Part#'s. Each Part#
uses up to 2 LotID's(Front and Back) but the Front LotID
is what affects the Part#'s performance.

- each end product comprises several elements? Yes
Correct?

What is the primary key of that table? Actually I have a

autonumber CompID PK as well as a multi-field index
between LotID and Element.
What does one row in that table, represent?
Row1: Auto#,LotID112233,Al,60
Row2: Auto#,LotID112233,Cu,40
The concatenated result would be LotID112233,AlCu,60:40

What does one row in that table, represent? The combined

elements from above are mixed with a source element to
produce the final combination which will be used in a part.

What is the primary key of that table? LotID is actually

the PK
What does one row in that table,represent? (I don't
understand: "characterization of LotID's after
manufacturing". You need to describe the table in terms of
the entitites you

have mentioned so far: parts, tests, elements, end

products, etc.) The combined/processed elements are tested
(characterized) before being used in Parts. Each row will
contain the results for a LotID

TC · Apr 30, 2004

Hi AAron

Sorry, but I still don't clearly understand the data that you're dealing
with.

Let's get the data descriptions correct, before we proceed. This may seem
picky, but it is actually the most important step in "data modelling". A
clear understanding of the data entities must come first - then the actual
design of the tables.

Is this correct:

- The research facility handles ELEMENTS, END PRODUCTS, and PARTS.

- I assume that each ELEMENT has a unique ELEMENTID? If not, how do you
distinguish different ELEMENTs?

- Two or more ELEMENTS are mixed together. The result is called a LOT. Each
LOT has a unique LOTID. There is no case where two LOTs have the same LOTID.

- Each LOT is PROCESSED to form an END-PRODUCT.

- Is there only one LOT per END PRODUCT, or could you process several LOTs
into a sinbgle END PRODUCT?

- I assume that each END PRODUCT has an ENDPRODUCTID? If not, how do you
distinguish different END PRODUCTs?

- Each END PRODUCT is subject to one or more TESTS.

- I assume that each TEST has a unique TESTID. If not, how do you
distinguish different TESTSs?

- Could one TEST handle several END PRODUCTs simultaneously?

Aaron, you may feel I am going backwards with my advice to you! (We do not
seem to be getting down into the nitty gritty of that table structures, at
all.) But the reason for this, is that you have not yet clearly described
the data that you need to manage. So IMO, it is premature to comment on the
table structures.

Unfortunately I'll be busy for the next 2 days, & have little or no time on
the net. But I'm happy to get back to this as soon as I return. Or maybe
someone else will jump in & help.

Cheers,
TC

Aaron · Apr 30, 2004

-----Original Message (edited w/reply*)-----
Hi AAron

Sorry, but I still don't clearly understand the data that you're dealing
with.

Let's get the data descriptions correct, before we proceed. This may seem
picky, but it is actually the most important step in "data modelling". A
clear understanding of the data entities must come first - then the actual
design of the tables.

Is this correct:

- The research facility handles ELEMENTS, END PRODUCTS, and PARTS.

- I assume that each ELEMENT has a unique ELEMENTID? If not, how do you
distinguish different ELEMENTs? *Yes I have a table of

Elements: Autonumber ID and ElementName

- Two or more ELEMENTS are mixed together. The result is called a LOT. Each
LOT has a unique LOTID. There is no case where two LOTs have the same LOTID. *Correct

- Each LOT is PROCESSED to form an END-PRODUCT.

- Is there only one LOT per END PRODUCT, or could you process several LOTs
into a sinbgle END PRODUCT? *One lot per END PRODUCT

- I assume that each END PRODUCT has an ENDPRODUCTID? If not, how do you
distinguish different END PRODUCTs? *The LOTID assigned

at element composition is the same as the ID for the END
PRODUCT. The reason for the different table is because
since it can have more than 1 element and I need to
capture the amount/ratio of each Element used, I thought
it best to have the separate table. If you have another
suggestion please let me know.

- Each END PRODUCT is subject to one or more TESTS.

- I assume that each TEST has a unique TESTID. If not, how do you
distinguish different TESTSs? *"Tests" are unique fields

such as ParticleSize, ECvalue, FinalMass. Again, it's more
of a characterization of the mixture of elements.

- Could one TEST handle several END PRODUCTs

simultaneously? *The Products are tested/characterized
independently but they all receive the same "tests".

Aaron, you may feel I am going backwards with my advice to you! (We do not
seem to be getting down into the nitty gritty of that table structures, at
all.) But the reason for this, is that you have not yet clearly described
the data that you need to manage. So IMO, it is premature to comment on the
table structures.

Unfortunately I'll be busy for the next 2 days, & have little or no time on
the net. But I'm happy to get back to this as soon as I return. Or maybe
someone else will jump in & help.
Cheers,
TC

*I appreciate the assistance thus far and I understand you
need to know the data.
--------------------------------------------------------

TC · May 3, 2004

Ok. Now we have the following clear description of your needs:

"The research facility handles ELEMENTS, LOTS, END PRODUCTS and TESTS. Two
or more ELEMENTS are mixed together, in various amounts or proportions, into
a LOT. Each LOT is PROCESSED to form an END-PRODUCT. (There is just one LOT
per END PRODUCT.) The END PRODUCT is then subject to various TESTS."

Here is the correct table structure to meet those needs (excluding TESTS).
Obviously your own table & field names may vary from mine - but there should
be a 1-to-1 correspondence between your tables, and these:

tblElement - one row for each ELEMENT known to the system.
ElementID (PK)
ElementName
- any other attributes of the element as a whole; eg. toxicity.

tblLot - one row for each LOT known to the system.
LotID (PK)
Processed - a yes/no flag to indicate whether this LOT has been
processed into an END PRODUCT yet.
- any other attributes of the lot as a whole; eg. date created.

tblLotElement - one row for each ELEMENT in each LOT.
LotID ( composite )
ElementID ( primary key )
- a field to give the amount or proportion of that ELEMENT in that LOT.

I haven't included TESTS, because I still do not understand your description
of those.

Aaron, I hope this helps. Note that I designed the table structures above,
in about 1 minute (literally), >once I understood your data requirements<.
That first "understanding" stage, is where most of the time lies. >Do not
try to design the tables during that stage<. The trick is to get an accurate
understanding of the data requirements, >before you start designing the
tables<.

HTH,
TC

Aaron · May 4, 2004

Thanks again. So far your tables line up with what I was
thinking of doing. I'm thinking now that the TEST fields
could just be in tblLot instead of a separate table
because they are characteristics of LotID such as
ElectricalPerformance, AvgParticleSize, etc. ????

TC · May 4, 2004

That sounds good. If the TEST fields are an attribute or characteristic of
the LOT as a whole, then they would go in tblLot. If it only made sense to
enter those fields >after< a lot had been processed into an end-product, the
relevant form(s) could check the Processed flag (in tblLot) to enforce that
restriction. For example, the Form_Current event could enable or disable the
TEST field controls, depending on whether the current tblLot record had or
had not been processed yet, respectively.

Good luck!
TC