Professional and Career Topics

One thing I have found that is important to impart to junior staff is the concept of uncertainty. Often in reports, I encourage them to round numbers to a certain extent. For example, instead of telling a client that a given model predicts a reduction of 81.52 MG of flow, state that is it "on the order of 80 MG" or something similar. This is an area where I feel I could continue to improve myself, in terms of understanding how much to round based on the type of modeling performed. 

I also recognize that there are jobs in this field where high degrees of precision are actually required, but it is important to understand when that precision is real and when it is an artifact of the calculations you have performed. 

Does anyone else have other examples of either ways they handle significant figures in communication results to others? Or perhaps any useful terminology they employ in these forms of communication? 

------------------------------
Christopher Seigel P.E., M.ASCE
Civil Engineer
------------------------------

Read about the operation of a slide rule.  Received my engineering education in the 1960s using a top-of-the line engineering slide rule and started work immediately thereafter.  Until the early 1970s a slide rule was used for the vast majority of engineering calculations.  Everything to do with a slide rule is about significant figures (typically 3 significant digits on the "standard length" 10" slide rule).  As you mention, emphasize the difference between "precision" an "accuracy".  Here is a simple example:

52.631 x 7.286 = 383.469466   ...Very precise, but how to do this on a slide rule?

By necessity, round off to three significant digits (both input and output)

52.6 x 7.29 = 383   ... intuition may say that this rounding will introduce significant inaccuracy.  However, 383 compared with 383.469466 is 99.88% accurate.

Bottom line, be as precise as possible during the calculation, then round off the end result, as you are suggesting.

------------------------------
Robert Higgins P.E., Life Member ASCE, Life Member ASME, Retired Member AISC
------------------------------

When I was a summer intern, I was asked to label pipe lengths on a construction document. I measured the lines and put what CAD gave as the output. My boss called me in when he was reviewing the plans to ask me if I thought a contractor was going to order or cut a pipe to the thousandth of a foot accuracy.  (He did it in a professional way as a teaching moment.)

Up until then, I was used to science and math classes where more significant figures meant more accuracy. I hadn't really stopped to think about how these numbers were being used. I now try to communicate that same concept to drafters and engineers as I review their plans. We should think about the plans from multiple perspectives based on how the plans are used. 

Fair or not, when someone using the plans in the field can tell that you didn't think about the real world when putting together the plans, they're less likely to take you seriously. Since successful projects often hinge on how well everyone works together, we should do our best to give the information they actually need without compromising the integrity of our calculations.

------------------------------
Heidi C. Wallace, P.E., M.ASCE
Tulsa, OK
------------------------------

Heidi and others who responded:

I think your supervisor's teaching moment is one you will never forget. Glad he did so in a professional manner. When I get tests and homework solutions that show 3 or 4 significant figures to the right of the decimal, it is annoying. So early in my semester, I usually ask students if they understand the difference between accuracy and precision. Most have a rather muddy notion of the difference. I then define accuracy as absolute nearness to truth.  Precision is the degree of care in a measurement or observation. The important point is that we can use very precise measurement tools, but they must be calibrated against truth. And darts, horseshoes, and hand grenades are not very precise instruments but can sometimes produce results very near absolute truth.  :-)

Dave Bennett, PhD, PE

Professor of Practice

Richard A. Rula School of Civil and Environmental Engineering

Mississippi State University

------------------------------
Robert Bennett P.E., BC.PLW, M.ASCE
President, CEO
Richard A. Rula School Civil/Enviromental Eng (aka
Mississippi State MS
------------------------------

Original Message:
Sent: 09-04-2025 09:47 AM
From: Heidi Wallace
Subject: Significant Figures Aren't Significant

When I was a summer intern, I was asked to label pipe lengths on a construction document. I measured the lines and put what CAD gave as the output. My boss called me in when he was reviewing the plans to ask me if I thought a contractor was going to order or cut a pipe to the thousandth of a foot accuracy.  (He did it in a professional way as a teaching moment.)

Up until then, I was used to science and math classes where more significant figures meant more accuracy. I hadn't really stopped to think about how these numbers were being used. I now try to communicate that same concept to drafters and engineers as I review their plans. We should think about the plans from multiple perspectives based on how the plans are used. 

Fair or not, when someone using the plans in the field can tell that you didn't think about the real world when putting together the plans, they're less likely to take you seriously. Since successful projects often hinge on how well everyone works together, we should do our best to give the information they actually need without compromising the integrity of our calculations.

------------------------------
Heidi C. Wallace, P.E., M.ASCE
Tulsa, OK
------------------------------

Dear colleagues,

Building on the excellent points raised about uncertainty and communication, I'd like to share a practical experience from a structural analysis course I taught, where students explored multiple approaches to solving the same problem.

One case involved applying the moment distribution method in Excel. A conventional, conservative use of this method often yields a safe result, but it can obscure subtle shifts in structural behavior. By deliberately increasing the internal precision (e.g., retaining more decimal places through iterative cycles), we uncovered outcomes that were not only more accurate; but also revealed opportunities for more efficient design results, comparable to those produced by advanced analysis software.

This level of precision is not a computational artifact; it's a gateway to the model's full predictive power. The key takeaway, which echoes your insights, is that we must be masters of our tools. We need to discern when internal precision is essential for rigorous analysis, and just as importantly, when and how to round results for effective communication with clients, who often care more about scale than decimal detail.

Thank you for this thoughtful and engaging discussion.

Warm regards, 

Abubakr Elfatih Ahmed Gameil

------------------------------
Abubakr Gameil, R. ENG, M. ASCE®️, SEI Member
Chairman & Director General
Almanassa Engineering International Co. Ltd
Khartoum, Sudan
------------------------------

Great question, Chris.

In reviewing papers and technical reports, I find authors frequently forget that a calculated output should be no more precise (number of significant figures) than the precision of the input data. If the input data have only two significant digits, the output can have only two.

Numerical modeling is more complicated. Models will, of course, output results with many digits/decimal places. However, the results are sensible only to the degree that we can reasonably measure and model the quantities of interest. For example, measurements of water level in a river can only be accurate to within  a few cm (despite instrument makers' claims) and the datum plane may be off by a significant fraction of a meter. Then, the model's accuracy must be validated by comparison with those field measurement and deviations of several dm are common. Such a model's results should be reported only to the nearest dm AND with uncertainty bounds of plus or minus "X dm" with the X based on a statistical analysis of the validation results.

Bill Mc

------------------------------
William McAnally Ph.D., P.E., BC.CE, BC.NE, F.ASCE
ENGINEER
Columbus MS
------------------------------

Interesting topic Christopher – and discussion contributions are great.

Robert Higgins' . . . Bottom line, be as precise as possible during the calculation, then round off the end result . . . captures the real-world approach that defines the activities of practicing engineers. The same is shared by Heidi - her transformative experience from the ideal student life of seeing things – to the practicality of civil-engineering-world of doing things.

As engineering involves numerous calculations – accumulation of imprecision is bound to occur when truncation of numbers are prematurely done. As in 2. Engineering, Civil Engineering and Coastal Engineering; 2.1 Definitions of Civil Engineering on Our Seashore . . . Engineering has always been and will always be – the practice of finding ways to derive workable and applicable solutions . . . And approximation and truncation at the appropriate level and phase – are part of the game to achieve the workable and applicable solutions.

Even then, uncertainty is unavoidable. As in Uncertainty and Risk – apart from the random uncertainty we are mostly familiar with – embedment of systematic uncertainties accompany all methods, tools, measurements and models. Therefore – as discussed in another thread, the question of implanting redundancy in designs becomes so important.

Accuracy and precision – appear into discussions often. Although used interchangeably – an old NIST Article clarifies the difference lucidly. Quoted frequently, the article highlights the confidence level in terms of 'how good' and whether 'good enough'.

A civil engineer's world revolves around answering these two questions to develop the right confidence level he or she needs.

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

Dr. Dilip K Barua, Ph.D

Website Links and Profile

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